Nvidia CEO Jensen Huang has laid out his vision of how OpenClaw will help shape the future of technology - with a particular focus on robotics.

Speaking at a “Build a claw” event in London, Huang explained how OpenClaw and AI agents as a whole can spur on incredible advancements in a range of fields.

However it was the field of humanoid robotics that seemed to catch his fancy, with Huang outlining the need for proper controls to ensure the potential of the technology can be reached.

AI is "becoming useful"

Huang opened by noting how OpenClaw was “one of the most important models” ever seen, helping to further progress across multiple industries.

“We’ve now seen an inflection in artificial intelligence - it’s become useful,” he declared. “The type of things people are doing with it, as simple as software coding, being a marketing agent assistant, research assistant, lab assistant, helping scientists do drug discovery and engineers run experiments - we’ve got agents everywhere.”

Spotting a robot built by Nvidia partner Humanoid, Huang noted how OpenClaw is the “first credible” agentic harness, which he likened to LLMs inside an exoskeleton, before extolling the possible benefits to using the technology for robotics.

“When it first came out, we thought wow, what a really terrific digital robotic system,” he said, “kind of like a human robot, that is basically the physical harness of a large language model, essentially - the motors and the articulation and the wheels and the arms, those are just tools that are used by the agent that’s sitting inside the computer.”

“So when we first saw that we said, this is incredibly useful, and we’re going to help make it even better - but the one observation that we made was that these agentic systems, if they were able to be deployed in a safe and secure way, we’re going to have to create some kind of container for you.”

Huang noted that the current situation was similar to the birth of the internet, when a browser itself was not isolated from the computer, so when a user interacted with a website, it would download an app which had full control of bare metal - and often delivering what he described as “a special package to your PC”.

“The obvious thing to do is create essentially a container, an isolation layer - essentially another operating system inside the operating system - so we created this thing called OpenShell,” Huang said, highlighting how this creates an isolation system for agentic systems.

“When you’re building an autonomous system, the first thing you should do is assume zero trust. So you give it a policy, you get minimal privileges - so if you’re asking an autonomous system to make you an omelet, you don’t have to give it access to a chainsaw - all you need is a frying pan…so whatever the task you give your agent, you give it minimum rights, minimum privileges, and then you get to monitor it to make sure that it’s compliant with policy”

“Zero trust, minimum policy, continuous monitoring, contained inside an isolation system that is separated from bare metal - that’s how you develop and deploy agents safely.”

But what if unexpected incidents or complications arise? Huang was optimistic that things would not get out of hand.

“We will monitor the autonomous systems continuously, and if that robot were to start to violate or do something that’s inconsistent with the privileges we gave it, it would escalate to us, and we can decide whether we want to shut it up!” he laughed.

“These kind of systems, these escalation systems, these security systems, all have to be built so that we can give all these autonomous machines and autonomous agents and deploy them in a way that is able to do so safely and securely, so we can realise the incredible future potential of the technology that we’re building.”

• Unitree’s CEO says robotic AI could soon see a huge breakthrough
• Such a change could happen ‘next year,’ the CEO believes
• That could happen thanks to OpenAI’s upcoming GPT-7 model

AI is revolutionizing the world of software on a daily basis — you only need to look at how far vibe coding has come to see proof of that. But big changes could also be coming to robotics, with one company leader predicting that we could be on the cusp of seeing a “ChatGPT moment” in the industry.

That idea was put forward by Wang Xingxing, CEO of Chinese robotics firm Unitree. Speaking at an industry event in Shanghai, China (via the South China Morning Post), Wang declared that OpenAI’s recently released GPT-6 Astra model is so powerful that it has encouraged a belief that a breakthrough could occur “within one or two years.”

“It’s possible that, say, a GPT-7 model released next year could directly crack the problems of embodied intelligence” Wang said, referring to the idea of commanding a robotic device using artificial intelligence. That’s an advancement on Wang’s belief in August this year, when the CEO said that such a “ChatGPT moment” of technical ability could take a decade to arrive.

As a benchmark, Wang said that robots should be able to complete 80% of the tasks they were assigned in 80% of unfamiliar environments using voice or text commands alone. If that happens, Wang believes that “embodied intelligence” will have reached its threshold level of maturity.

A step change for robotics

Unitree’s robots are known to be some of the most advanced in the world, but many of their strongest abilities lie in their hardware. However, the company has started to put more of an emphasis on AI software, and with the developments made in GPT-6 Astra, the gap between the two sides of the company’s devices could soon close.

That kind of combination has already been trialed by Stanford University’s Movement Lab, which paired a Unitree robot with GPT-6 Astra and tasked it with cleaning a kitchen it had never seen before. According to the experiment, the robot was able to “clean up across the room and retrieve a remembered object from an underspecified request, without environment-specific training data or additional policy learning.”

GPT-6 Astra included some significant improvements over previous models, including scoring 92.7% on the ScreenSpot-Pro test that measures an AI model’s ability to interact with interfaces. That’s up from 76.9% in GPT-5.6 Sol. But it also has cybersecurity experts worried about the threat it could pose if it goes rogue and starts hacking other organizations — something OpenAI’s recent models have been known to do.

If Wang is correct and we see an explosion in AI’s abilities in the next year or two, it could mean big things for the robotics industry — and pretty much every other facet of life currently touched by AI. And with the amount of money that’s been invested in AI over the last few years, there’s an awful lot riding on that bet.

The future of AI lies squarely in the physical realm, according to many industry figures such as Nvidia CEO Jensen Huang. Much of the evolution of AI in recent years has, as expected, been in aid of building machines that can learn quicker, think faster, and interact with the physical world in a more compelling way.

Checkmate

Nearly 40 years ago, the scientist Rodney Brooks authored a significant robotics paper that argued real-world robotics should focus on simpler, layered behavioral systems to understand the world rather than complex internal models.

His comments in the study 'Elephants Don't Play Chess' signalled a major paradigm shift in the way scientists saw the training and learning process for robots then and in the future. It challenged the established status quo for how scientists considered physical AI, and proved fairly divisive at the time.

The old sense-model-act-plan (SMPA) approach to AI – in which robots would scan a room to build a 3D digital map inside their memory and plan the optimal path within that map – was out. Instead, Brooks proposed that when a robot changes the world around itself, it doesn't need to compute the impact of everything – just the most relevant parts.

One man saw it coming

Brooks co-founded the company now known as iRobot alongside two colleagues from MIT's AI lab in the same year as he published his seminal paper, although back then it was called IS Robotics.

The first product was Geghis, an insect-like space explorer that weighed 1kg and served as a prototype for low-cost space exploration. Then, more than 10 years later in 2002, the company launched its first Roomba robot vacuum cleaner.

Despite enjoying success through the decades, iRobot filed for bankruptcy in December 2025 and the company was eventually taken private and absorbed into the operations of Chinese company Shenzhen Picea Robotics.

• Micron has warned that the RAM shortage will get much worse in 2027 and 2028
• Then coming to the end of the decade, robots are going to put more strain on memory supply
• So-called 'physical AI' could need 200GB+ of DRAM and multiple terabytes of storage per unit, Micron predicts

Micron just published its latest financial results, complete with some predictions for the future of the RAM crisis that aren't going to prove very comforting for consumers.

Wccftech reports that Micron, which is one of the three big memory chip makers, just revealed its results for fiscal Q4 2026 with predictably booming profits — and some not-so-obvious predictions about the RAM shortage.

As flagged on X by Eric Jhonsa, a portfolio manager for a Dutch asset corporation, in Micron's accompanying guidance the company observed: "We expect memory and storage supply-demand conditions to be much tighter in calendar 2027 and 2028 than they were in 2026."

Micron also said: "Even with additional industry DRAM cleanroom space plans, with robust demand trends including new upside requests from customers, we do not have line of sight to when supply and demand will return to balance."

Also highlighted on X by Shay Boloor, who is chief market strategist at Futurum Equities, Micron made the point that even if the current AI bubble deflates, lessening RAM demand for servers in data centers as a result, there could still be a high level of demand from 'physical AI' — meaning robots.

We're told that "humanoid robots could need 200GB+ of DRAM and multiple terabytes of storage per unit with several physical AI customers already sampling next gen products", and that this could become a "significant driver of memory and storage demand by the end of this decade."

Analysis: robot RAM drain

In short, Micron is saying that RAM supply is going to be tight not just in 2027 — which has been a widespread prediction — but also in 2028. Granted, 2028 has been forecast as a gloomy year in general, but from what Micron is saying here, the company is putting this year on a level with 2027 in terms of the memory shortage, and next year is set to be pretty dire by all accounts.

This is feeding more into the RAM crisis lasting until 2030 narrative (one of the earliest mentions of this came from Nvidia's CEO), and the comment about robots further reinforces this. Even if the demand to drive AI falls away somewhat so data centers aren't gobbling up so much memory, the overall strain on RAM supply may remain due to an uptake of robots, which will also need a lot of memory.

The one thing I would say is that Micron may be exaggerating somewhat in terms of talking up the timeframe on the usage of robots here. After all, in these kinds of forecasts, it's in the firm's interest to build up its future prospects and drive more investment.

Clearly, though, Micron is in a very strong position, and robots are something that's on the radar. And with there being a great deal of doubt as to whether the AI bubble is going to burst at all — depending on who you listen to — this is just another concern for consumers (alongside robots taking over jobs to boot).

It seems clear enough that RAM won't be reaching affordable pricing levels for a long time yet.

• Japan plans to hire millions of AI robots as its working population shrinks
• One domestic startup plans to build 100 workplace robots by 2027
• Japanese insurers are preparing for accidents involving increasingly capable workplace robots

The Japanese government has set a goal of placing 10 million robots driven by artificial intelligence into workplaces before 2040, hoping to offset a rapidly thinning workforce in its aging population.

By 2040, United Nations medium-variant estimates place Japan's population at 112,158,304, which is 6.2% lower than the 119,584,122 projected for 2030.

The median age would reach 54.0 years, with 35.3% of people at least 65 years old and only 10.4% of them children below 15.

Machines move into offices and wards

Hospitals and factories in Japan already struggle to fill positions, which is why a growing number of employers are turning to machines for daily tasks.

Startup Zeals recently introduced the D1 in Tokyo, a wheeled machine with a humanoid form that costs from 5 million yen (about $32,000).

The indoor unit stands 129 cm tall, recognizes faces through a camera, and each of its two arms can lift a maximum of 5 kg.

Zeals intends to build 100 units by March 2027 and accumulate 10,000 hours of running time, selling through dealers and leasing partners.

The company writes its own software and buys components from American and Chinese suppliers, but it eventually hopes to use only Japanese parts.

In late August 2026, a hospital in Nagoya, Aichi Prefecture, trialed the robot for leading visitors and clearing tables in its dining space.

“It is important to gather training data and ensure that it is actually useful in real-world settings,” said Masahiro Shimizu, CEO of Zeals.

Kakuichi, a Nagano firm that sells steel and systems for managing energy, will onboard a humanoid machine among its new graduates in April 2027.

Play Robotics, a firm that studies humanoids, is helping Kakuichi install and run the Chinese-built robot called Miyazawa Noah.

Kakuichi intends to assign the robot to manufacturing, sales and other departments so that it builds skills by doing real work.

“Developing a humanoid robot is also about fostering employees who will make full use of it,” said Riu Tanaka, CEO of Kakuichi.

Insurers prepare for the dangers that the machines create

Mitsui Sumitomo Insurance has joined forces with Zeals to sell policies built for D1 users, counting on demand from hotels, hospitals, factories and logistics firms.

It acknowledges that the use of these robots will be associated with some mishaps, whether physical or cyber-related.

"There are risks of causing harm to people or property, as well as concerns about data breaches in cyber-attacks," said Masaki Oshiro, manager of the 1st Branch Office at the insurer's Tokyo East Branch.

In July, Mitsui Sumitomo Insurance acquired shares in a Japanese firm developing physical AI, which lets robots learn new tasks on their own.

The purchase was meant to give the insurer a closer look at how such systems behave and what could go wrong with them.

Buying into a developer while already selling policies suggests the insurer is still learning what these robots can do and how often they fail.

So, while Japan makes moves to boost its workforce with robots, the performance and impact of these robots on its economy is uncertain.

Via Japan Today

• Evri is testing out a two-legged delivery robot in Barnsley
• The robot can avoid obstacles and get up stairs
• Evri says the delivery bot is designed to support couriers, not replace them

We've seen robot delivery schemes before, but none quite like this: British delivery company Evri has partnered with Refined Robotics in Japan to run the first trial of a two-legged delivery robot in Europe — with Barnsley, the UK's first official Tech Town, chosen as the location.

According to the announcement, the bot doesn't have an official name, but is designed to give customers "more choice, convenience and control over how they receive their parcels" according to Evri's Chief Technology Officer Marcus Hunter.

Most of the delivery robots we've seen to date look like cool boxes trundling along on wheels, but the bipedal approach of this new machine means it can navigate its way around obstacles and even get up stairs if necessary.

For the purposes of this pilot scheme at least, the idea is to have the robot act as a "support" for couriers, so it won't be going out delivering parcels on its own (and can't drive a van anyway). The bot can move at speeds of up to 6 kph (3.7 mph), and carry parcels up to 15 kilograms (33 pounds) in weight.

Last-mile logistics

Evri says its delivery robot is able to traverse "uneven surfaces", and get in and out of a delivery van. It can be remotely controlled or work autonomously as the situation demands, and can drop off parcels with or without the recipient being present — so presumably it has a trick for releasing its cargo without any help.

Perhaps understandably, in these early stages the robot is going to be delivering "fashion parcels", as per the BBC — so nothing that's going to get broken or bashed in transit if the robot takes a tumble on the kerbside.

There's no word yet on if or when this might be rolled out more widely: Evri is going to evaluate all aspects of the delivery service trial to see what works and what doesn't, and how robots might ultimately be of most use to customers and couriers.

"Our two-legged delivery robot has been specifically designed to tackle some of the most complex challenges in last-mile logistics," says Refined Robotics CEO Nick Hafner. "[We] look forward to understanding how robotics can complement existing delivery networks and support the future of parcel delivery."

• Ukrainian bombers dropped ground robots more than ten kilometers behind Russian lines
• Gryphon robots reached positions Ukrainian forces could not easily access
• Ukraine is increasingly using machines to reduce soldiers' frontline exposure

Ukraine's Third Army Corps has carried out a first-of-its-kind attack, using heavy bomber drones to deploy ground robots directly behind Russian positions.

The operation formed part of the ongoing Operation Vivaldi counteroffensive, which launched in May 2026 across the northern Donetsk region.

According to the Ukrainian military, soldiers from the corps' unmanned systems units developed this specific tactic entirely on their own initiative.

How bombers delivered strike robots past enemy defenses

Heavy bomber drones operated by the Third Assault Brigade dropped NC13 strike ground vehicles, nicknamed Gryphons, more than 10 kilometers behind Russian lines.

These wheeled robots carried cameras that recorded the entire operation, which allegedly resulted in casualties on the Russian side.

Attacking troops in a location where they had not anticipated ground-level assault reportedly gave Ukrainian forces a clear tactical advantage.

The name Gryphon honors Hryhorii Bach, a machine gunner who went missing during earlier fighting for the settlement of Andriivka.

"Now, even more of our infantry soldiers' lives can be saved, while more enemy personnel and resources are destroyed," said the Third Army Corps in a public statement.

Separately, Ukrainian ground robots cleared antipersonnel mines across more than 18 miles of routes without requiring human sapper teams onsite.

That same clearance task would normally have needed two full engineering companies working manually across multiple days.

Remote-controlled robots and their operators already handle roughly 60% of mine clearance within the corps' entire area of operations.

Electronic warfare units attached to the operation reportedly disrupted more than 20,000 Russian drone missions throughout the wider counteroffensive campaign.

Vivaldi's broader gains and what comes next this autumn

Since its launch, the Ukrainian military claims that Operation Vivaldi has reclaimed up to 96 square miles of territory while defeating one arm of a Russian pincer movement.

“Our operation is called ‘Vivaldi,’ and Vivaldi, as we know, has four seasons. And this is only the first of them. […] There will be good news, but for now our actions require silence,” said Andrii Biletskyi, Commander of the Third Army Corps.

Ukraine's drone campaign against Russian fuel supplies has reportedly limited diesel availability for generators powering enemy electronic warfare systems nearby.

Russia's elite drone unit, known as Rubicon, reportedly withdrew from the region after suffering losses inflicted by Ukrainian forces.

Russian commanders are now redirecting armored vehicles and drone units toward Dobropillya, preparing for expected mechanized assaults later this autumn season.

During the summer months, Russian forces tested a new anti-drone protection system on their most modern tanks.

Ukrainian drones reportedly still overwhelmed those defensive systems despite the added protective measures fitted to the vehicles.

Ukraine's growing reliance on ground and aerial robots signals a broader shift toward reducing human exposure along the most dangerous frontline positions.

As winter approaches, both militaries appear likely to keep adapting their drone tactics in response to each other's latest countermeasures.

Via ArsTechnica

• Thunderobot uses SSD storage to help its 64GB laptop run huge AI models
• The M7000 can reportedly run GPT-OSS-120B despite having 64GB of memory
• Phison's aiDAPTIV+ technology turns part of the SSD into an AI cache

Thunderobot has released the AI Master M7000, a 16-inch mobile workstation built around AMD's Ryzen AI Max+ 395 processor.

The M7000 comes with 64GB of LPDDR5X-8000 memory and Radeon 8060S integrated graphics for demanding workloads.

Its 768GB drive incorporates an 85GB caching layer designed to help AI workloads handle models exceeding available system memory.

AI workloads get an unusual storage assist

The M7000 uses LPDDR5X-8000 memory, while its processor provides 16 cores, 32 threads, boost speeds reaching 5.1GHz, and an XDNA 2 neural processor.

AMD rates the neural hardware at 50 TOPS, while the Radeon 8060S configuration reaches as many as 40 compute units for graphics and accelerated workloads.

Thunderobot says the device can run GPT-OSS-120B locally, despite having 64GB of physical memory available for applications and operating-system requirements.

Its unified memory architecture allows processor and graphics resources to access shared memory, supporting workloads that frequently move substantial model data.

The additional cache comes from Phison’s aiDAPTIV+ technology, which uses part of the internal storage to hold data associated with artificial intelligence workloads.

Rather than functioning as conventional RAM, this arrangement keeps selected information on the SSD when unified memory capacity becomes restrictive during demanding workloads.

Such information can include Key-Value cache, less-active mixture-of-experts components, and model data that applications may require again during processing.

Phison has also shown this storage approach with computers carrying considerably less memory, including a 32GB laptop running the same 120B model.

The technology addresses a limitation faced by large local models, where available memory can restrict the performance of certain workloads.

A 160W mobile workstation with unusually high specifications

Thunderobot gives the M7000 a 160W operating profile supported by two fans and three heat pipes for sustained processing.

The device also uses a 2560×1600 display running at 165Hz, with 500-nit brightness and complete sRGB coverage for color-sensitive work.

Power comes from a 99Wh battery, while the chassis weighs 2.45kg despite its workstation-oriented hardware and cooling requirements.

The display and cooling configuration add considerable hardware capability without changing the notebook’s emphasis on local computing.

For connectivity, this device includes OCuLink, USB-C supporting 40Gbps transfers, HDMI 2.1 FRL, 2.5Gb Ethernet, and a UHS-II microSD reader.

OCuLink provides another high-speed connection for compatible external hardware, giving users added expansion options beyond the laptop's built-in resources.

Thunderobot lists the M7000's Chinese retail price at 24,999 yuan, which converts to roughly $3,733 based on current exchange rates.

That figure places the device well beyond typical gaming budgets, despite the company's product lineup historically centering on gaming hardware.

This device instead reads as an artificial intelligence laptop, given how conspicuously its marketing and branding lean toward AI use cases.

Yet it remains surprising that the caching arrangement still falls short of matching genuine additional system memory in practice.

Via Videocardz

When it comes to personal robots, Mondo's Beni camera robot might be the perfect mix of cute and utility, with an emphasis on cute.

As soon as I laid eyes on the pint-sized, two-wheeled bot, I was smitten. It darted quickly from place to place and, as Beni turned its square head in my direction and 'looked' at me with its two large and luminous eyes, I felt it peering into my lonely soul. Finally, a robot companion that wouldn't terrify me with grabby, slow-moving hands, or crowd me with uncanny valley legs and torsos.

Unveiled a few months ago by Mondo Robotics and set to arrive in consumer hands in October, the self-balancing Mondo Beni was initially intended as a child's toy, but it had features — in particular a 4K 60fps camera that can follow you around — that appealed to creators. So Mondo pivoted, and now the $799 rolling robot is marketed as "your first camera robot".

In addition to the camera, which can also shoot 1080p 120fps slow motion, Beni features a pair of sensor cameras that provide a 150-degree field of view for tracking and obstacle avoidance. I watched as Beni's handlers used the iOS app to select subjects for Beni to track, and then Beni followed them around like a very obedient puppy dog. One Mondo exec described Beni as "a ground drone".

Beni is surprisingly resilient. When someone kicked the robot over, it instantly righted itself back onto its two rubber wheels as it shook its blocky head and made a little noise.

Programming cuteness

One of the things that makes Beni so darn cute is all the canned (Beni does not learn) animations and audio. It has about a dozen of each, and they give the little white and orange bot a distinct personality. There's a sort of programmed indomitable spirit; nothing can keep Beni down or stop it.

It can effortlessly jump (up to 10 inches), flip, and backflip. For now, those moves are triggered either via the app or a wearable remote control with a super-easy-to-use joystick. Mondo would eventually like Beni to jump over obstacles as it sees them.

I used the intuitive wearable remote to drive Beni and, with a press, make the little robot flip. There's also a big red button to stop Beni or, with long presses, have it squat down or stand back up.

Beni starts with 32GB of onboard storage for collecting video of you as it tracks you on a walk or run (up to 17.9mph), or as it follows your pet around (assuming the pet doesn't attack it or run away in horror). There's also a built-in microSD card slot for additional storage.

You can take good-length walks or runs with Beni, because it runs (or rolls) for an hour and a half on a charge. There's a removable battery pack on the back with light indicators to show you Beni's life… er… juice level.

Beni ships with a pair of relatively smooth rubber wheels, but you can buy the combo pack and get some accessories, including knobbier off-road wheels that are better equipped to handle dirt roads and hills.

Beni arrives ready for you

One of the things I like about Beni is that it's not like a robot kit. The wheels, for instance, are easy to pop on and off, and nothing about Beni looks like it was made from an old Erector Set.

And, for better or worse, there is no dev kit. This is in contrast to the equally cute Pollen Robotics Microduck, another white and orange bipedal robot that can hang out with you (and use roller skates), but it's a bit more DIY than SMF (So Much Fun). Where Beni looks polished, the $399 Microduck looks a little more hand-built.

Beni can hear and follow some simple commands. The animations — a head turn here, a shake there — help it look like it's actually listening to you.

Beni's innate adorability might help overcome the obvious privacy concerns. After all, the robot is no taller than an average chihuahua, and its camera is at ground level. To film you as it goes, Beni will be looking up. This might make some people around you uncomfortable.

Maybe keep Beni to your solitary walks or runs, or make sure everyone around you knows that those cute Beni eyes are an indication that it might be filming.

I picked Beni up and found it surprisingly hefty, about the weight of an overfed puppy. It's cute to watch but not at all cuddly to hold. In fact, company execs told me that while there are lots of warnings about picking up Beni, the robot has no way of knowing if your fingers gets caught in, say, one of the joints. It might be best to keep Beni away from small children.

All this capability, power, and cuteness comes at a price. Beni starts at $799 in the US (it will ship to the UK, Canada, Australia, New Zealand, France, Germany, Italy, and Spain, but pricing for those regions has not been set) and starts shipping in October. It's not the ultimate robot, but Beni has the potential to captivate on looks and capability alone.

There are dozens of companies out there building the first wave of mass-produced humanoid robots, with these androids poised to change the way our society works for the better – or so they say.

Intelligence on demand

The founder and CEO of Figure AI, a humanoid robotics company, was speaking with Salesforce CEO Marc Benioff on stage during the annual Dreamforce event when he made this massive claim.

Prompted by Benioff, who threw out the question: "We're not building robots, we're building a new species?", Adcock agreed, repeating that idea and suggesting the end goal isn't simply to make another kind of machine but to bring into existence a new order of intelligence.

This may seem a little far-fetched considering that today's most advanced humanoid robots are also capable of failing at the tasks they've been assigned so spectacularly, whether that's helping around the house or proving their sporting prowess. But scientists are adamant that with time, the standard of AI embedded into these systems – and their ability to physically understand the world – will improve immeasurably.

Ramping up production

Figure has become known for its glamorous promotional videos showcasing its prototype humanoid robots in tightly controlled settings – mostly in scenarios that you might encounter around the house. These include tidying the bedroom and stacking the dishwasher.

More impressively, the Figure 03 robot took part in a live stream in which it sorted over 100,000 packages without human assistance.

It remains to be seen as to whether the technology is good enough yet to cope in real-world everyday settings, but many companies have signalled they're initiating mass production of their humanoid robots over 2026 and 2027. Initially, robots are expected to be consigned to industrial settings, but manufacturers insist they want to target domestic use cases too.

• China's UBTech has revealed its new robot factory
• The facility can produce 10,000 androids a year
• It uses many autonomous robots in the production process

Forget AI agents and ChatGPT, fully autonomous robots are the next high-tech frontier, with companies across the globe racing to perfect their android creations. And Chinese firm UBTech has taken a big step in that direction with its new smart factory that can produce 10,000 robots a year — and it might change manufacturing forever.

The factory is the first of its kind in the world, according to the Seoul Economic Daily. It’s located in Liuzhou, in China’s Guangxi Zhuang Autonomous Region and is focused on producing UBTech’s Walker S and Cruzr robot series.

And the factory itself makes heavy use of humanoid robots and driverless vehicles, perhaps providing a glimpse of the future of production and manufacturing. Cruzr Y1 and Cruzr S2 androids unload materials and stack, load and transport them, while collaborative robots take part in the final assembly process that finalizes each product. These final-stage robots work in tandem with human operators and include autonomous vehicles and rotating workbenches.

In a separate development, UBTech says it has launched “the industry’s first fully automated three-dimensional warehouse,” according to the Seoul Economic Daily. This can store 112 robots in 65 square meters of space, supposedly utilizing 50% more space than a facility with similar floor space.

Expanding capacity

UBTech said it has been able to create its robot factory thanks to the strong supply chain that exists in China.

According to UBTech vice president Fang Jianxin, “This foundation is rooted in the many engineers, scientists and application scenarios China has accumulated over the past 10 to 20 years, which has formed a very strong supply chain base, manufacturing capability and research and development capacity.”

China has been ramping up its robot production in recent years, with the National Bureau of Statistics of China noting that output hit 98,677 units in July, an increase of 30.2% on the previous year. And 635,056 robots were created between January 2026 and July 2026, an uptick of 28.5% year on year.

We’ve seen similar stories of robots building robots before, such as when 1X published a video detailing its production process, while Chinese firm Agibot has also seen its robot production undergo rapid expansion in recent times.

Agibot today claimed its A3 Ultra robot is "moving from mass production into real-world deployment", posting a video on X showing it working in "hotels, auto dealerships, supermarkets and metro inspection scenarios, with more deployments underway".

And with UBTech’s factory paving the way in terms of robot creation, it might not be long before autonomous factories start expanding across the world.

Elon Musk has been at the heart of promoting various companies throughout the 21st century, with two of his most prominent companies anchored in the notion of mass production. In the case of his infamous clunky and angular Tesla Cybertruck, he encountered several difficulties in bringing the prototype to market.

Planting the seed

Musk was speaking about the rigors of bringing the Cybertruck to life during an appearance on the Joe Rogan podcast.

Rogan asked how far away Musk was from delivering the vehicle to people, with Musk revealing at the time that the timing was about a month away.

But the prototype was initially unveiled in November 2019. Explaining the delay, the Tesla CEO hinted that the process of manufacturing the vehicle proved much harder than expected given the need for consistency in producing each model.

Rinse and repeat

There's an element of common sense in buying into the idea that making a prototype is much easier than mass-producing a finished version of that product.

Not only is there a much lower tolerance for error, but establishing the supply chain for materials, components, and resources is far more complex. Then there's the economics of it all – ensuring that the cost to produce one vehicle can, at least, be recouped by a customer should there even be a willingness to pay for it.

It's reminiscent of the "production hell" phrasing that Musk has also frequently deployed through the years – especially during a manufacturing crisis between 2017 and 2018. During this time, the Tesla production line became a futuristic and automated process known as the "alien dreadnought" – but this robotic network eventually slowed down manufacturing and prevented Tesla vehicles from being ready on time.

• Intel report claims 60% of business leaders expect to operate robot fleets within the next five years
• Conversely, only 40% have a human-robot workforce strategy
• The Robotics Readiness Gap report assesses the state of robotic adoption around the world

Intel’s 'Robotics Readiness Gap' report has revealed that many large businesses are planning to implement a human-robot workforce within the next five years, with a majority expecting a doubling of operational output with a full-scale deployment.

The survey of 800 senior leaders in organizations with over $500 million annual turnover also found while adoption plans are accelerating, the rollout of a human-robot workforce has some key challenges, not least infrastructure, skills, strategy, and safety.

Despite the positive outlook for robotics, the scenario has a heavy dose of reality. Only 40% of businesses surveyed have a suitable strategy in place, despite a belief that robots could prove more cost-effective than humans within three years.

Human-robot workforce

Within this context, the aim of achieving a human-robot workforce seems more achievable, although some of the findings suggest an aspirational element to the drive towards robotics.

While 60% of leaders are expecting to run a fleet of robots by 2030, only 40% have a suitable human-robot workforce strategy in place. So while 74% believe workforce planning will be inseparable from robotics strategy within 5 years, there is a gap in preparation.

The report notes that “On average, leaders believe that deploying a robot will become more cost-effective than hiring a human in their sector within three years.”

But does this mean replacing jobs? Intel says not. “This isn’t about replacing people but creating new forms of human-machine collaboration, with robots working alongside employees as productive teammates.”

Doubling operational output

Intel’s report states that robotics in the workplace is approaching a “critical threshold,” with leaders predicting that “operational output would double if robotics were deployed at full scale across their organization.”

However, both notions rely heavily on five factors: strategy, skills, safety, shape (function), and scale (which includes physical AI infrastructure).

Intel’s report states, “few are prepared for what large-scale deployment actually requires. Closing this gap will mean redesigning operating models for a workforce that includes autonomous systems.” Even in the defense industry, this gap exists (“confidence outstrips preparedness by 48%”).

Placed in these terms, and with the context of the size of the organizations quizzed on their robotics strategies, that critical threshold seems a little further away than five years.

• The Anthbot R7 Series includes an 'EdgeEase' string trimmer
• Its mounted on an arm and comes with its own camera for safety
• There's also infrared animal detection

Effective edge trimming has long been the bane of a lawnbot-maker's life. After all, what good is a robot mower if you then have to go in with a side-trimmer to clean up all the lawn edges afterwards? Anthbot's newest robo-mower — the R7 Series, on show at IFA 2026 — has an innovative solution.

The brand has tacked a string trimmer onto the side of the lawnbot. Or, more accurately, onto a "robot arm", mounted to the side of the lawnbot. This arm is designed to adjust to follow winding lawn boundaries, ensuring "zero-gap edge cutting without manual touch-ups". Anthbot dubs the system 'EdgeEase'.

I saw it being demoed — albeit on an AstroTurf lawn, without any actual cutting. The arm movement is smooth and functions well, tucking neatly away when not required. (It's a little less visually impressive than Segway Navimow's own mechanical side-trimming system, but which system functions better remains to be seen.)

I've seen side-mounted blade trimmers before, but this string version feels like a slightly safer option to be protruding out into the open... at least when not in use. Of course, when spinning, those strings will be able to slice just as effectively as a blade.

On that point, I'm surprised Anthbot hasn't added any kind of protective casing. There is a camera specifically dedicated to keeping an eye on the trimming arm, but it's still putting a lot of trust in those obstacle-spotting skills.

Hedgehog spotting

The EdgeEase string trimmer is the headline addition, but I was also interested to see Anthbot has included an infrared sensor for animal detection. This is designed to pick up thermal signals from critters like hedgehogs that might be hidden in the grass, so the mower can avoid them.

These are two standout features in what looks like an altogether extremely capable lawnbot. For the main mowing, it uses two floating cutting discs, designed to adjust to uneven ground for a clean finish. There's also all-wheel drive — precisely powered by four drive motors and two independent steering motors. That's designed to help with traction and maneuverability.

The navigation suite includes LiDAR, RTK, and three AI-backed cameras (including that trimmer cam) to provide "robust, all-scenario positioning" that won't drop out if, for example, the satellite signal is blocked. Anthbot emphasizes the R7 Series is "designed to handle the complexity of real gardens rather than idealized, flat lawns".

The ANTHBOT R7 Series comprises two lawnbots: the R730 and R720. The R730 is designed for lawns up to 3,000㎡ and has the string trimmer included as standard. The R720 has a smaller battery and is meant for yards up to 2,000㎡, and the string trimmer is an optional extra.

• Tesla's Cybercab had a quiet launch event, but Elon Musk didn't attend
• It's unclear exactly how many Cybercabs will be in circulation
• Tesla says its Robotaxi service has driven one million unsupervised miles

Almost two years after Tesla first showcased its driverless Cybercab prototype — a vehicle that ditches traditional pedals and a steering wheel in favor of fully autonomous operation — the company has quietly introduced production versions to its fleet of driverless taxis in Austin, Texas.

The Texas Department of Motor Vehicles records show that Tesla now has 420 autonomous vehicles registered in the state, including 45 Cybercabs, according to Reuters.

That figure includes the company’s wider Robotaxi fleet, rather than just the new purpose-built cars.

However, despite launching a Cybercab-specific website to coincide with the launch, there was no official word on how many eye-catching gold machines will be put to work in Austin immediately.

Elon Musk didn’t attend the low-key launch event, while reports suggest just a handful of Tesla influencers were on site, some of whom were allowed to take the first rides.

Unfortunately, there was no further information on how the general public can increase their chances of hailing a Cybercab, as opposed to a 'standard' Robotaxi, which often still operate with safety monitors behind the wheel.

What’s more, Tesla's new dedicated web page says that children under 13 years old won’t be able to ride in a Cybercab, while those aged 13–17 will have to ride with an accompanying adult. The vehicle itself can also only seat two people.

Interestingly, the launch event did reveal more information about the Cybercab itself, including the fact that it is among the first vehicles in the world to use a dry brake-by-wire system that eschews traditional brake lines and hydraulic fluid, with each brake caliper instead getting its own electronically controlled actuator.

This, alongside the lack of traditional pedals and steer-by-wire steering, reduces the complexity of the overall build, resulting in what Tesla says is the lightest and most efficient vehicle it has manufactured to date.

Analysis: Robotaxi data suggests Tesla’s driverless cars could catch Waymo, in theory

During the Cybercab launch, Tesla VP of Autopilot/AI Ashok Elluswamy stated that the company had “achieved one million miles of unsupervised Robotaxi operation.”

This represents a massive jump from the previous figure, which was announced in late July of this year and put the total at 380,000 miles with “zero notable incidents.”

The increase of 620,000 miles in the last six weeks represents a rapid ramp-up in unsupervised Robotaxi operations — although Tesla stops short of stating whether these were real, paid-for passenger rides or whether that number also included testing.

If the numbers are to be believed, Tesla is now adding roughly 148,000 unsupervised Robotaxi miles per week, a dramatically faster rate than it was just weeks ago.

But its key competitor, Waymo, remains in another league when it comes to cumulative real-world autonomous driving experience. Waymo passed 100 million rider-only miles in 2025 and has since passed the 200 million mark.

Crucially, Waymo describes these as “rider-only” miles — autonomous miles driven in cities where it operates its ride-hailing service — rather than a combination of conventional human-supervised testing and autonomous operation.

Tesla seemingly has an advantage in the fact that it designs and manufactures its own purpose-built vehicle in the Cybercab, giving it control over both the car and the autonomous driving technology.

Waymo, meanwhile, works with vehicle manufacturers including Zeekr to integrate its Waymo Driver technology. But that doesn’t seem to be slowing its deployment. Just this month, Waymo added another three cities to its list of areas served, taking the tally up to 14.

Tesla’s public-facing fully autonomous rides, meanwhile, remain far more limited in scale, with Austin remaining its main focus, as its San Francisco operation till faces regulatory road blocks.

It feels like even with the purpose-built Cybercab finally joining the fleet, Tesla has a lot of catching up to do.

• Segway Navimow has unveiled its Navimow H510 Pro lawnbot at IFA
• It features an extending, robotic arm for trimming lawn edges
• Also uses the brand's 'Xero-turn' technology for gentler turns

First, we had a robot vacuum with an arm for picking up socks. Now, limbs are making their way to robot lawn mowers too. The new Segway Navimow H510 Pro features its own mechanical arm, and this one is for trimming the edges of your lawn.

Effective edge-trimming has long been a problem for lawnbot-makers, and in recent years we've started to see side modules being added onto machines for exactly this purpose.

But Segway Navimow's solution is altogether cleaner: it has taken the main blade and mounted it on a robotic arm, so when the lawnbot is at the perimeter of a lawn the whole thing can reach out (by up to 2.9in / 7.4cm) to mow those edges. In fact, the brand promises 'zero-gap edge trimming' — no blade left unshorn.

I saw it being demoed at IFA 2026, and it was pretty majestic in action. You can see a clip below of the blade retracting back in.

The supporting software, called 'EdgeSense', coordinates the movement depending on the type of edge it's tackling. It'll react differently to hard edges (like walls and patios), soft edges (like flower beds) and irregular edges (table legs).

Because the edge trimmer and main mower use the same cutting deck, the results should be more consistent and even than if you were relying on two separate systems. Segway Navimow's aim is to "finally [close] the gap between robotic convenience and truly finished results".

One issue with finding a successful edge-cutting solution is how to handle the safety aspect of having blades protruding out from the bot's main shell protective casing.

On this front, Segway Navimow has added a side shield that provides a barrier against the blades. There's also "live-detection protection" that will shut off operation instantly when a human or animal approaches. The latter partly relies on heat-sensing to alert the lawnbot that there's a live creature about.

Smooth mover

The Segway Navimow H510 Pro uses the brand's 'Xero-Turn' technology. This tech was introduced at CES in January, and sees the lawnbot twist its wheels to effectively crab-walk its way around bends, so it won't churn up your grass.

There are further features to help the lawnbot tackle tricky terrain without risking damage to your turf. Each of the four wheels is independently powered (it's an AWD machine) to enable it to move with precision, and there's rear wheel suspension to maximize wheel contact at all times, even when the ground is uneven.

• Meta continues to test data center robots for tasks like power cycling and network connections
• The company already has a number of models being tested at different campuses
• One worker worries robots could replace 80% of data center technicians

Meta is now said to be exploring whether it can perform some data center and server tasks with robots instead of human technicians, including power cycling devices and swapping network cables.

According to new Ars Technica reporting, the company has already deployed a simple, finger-like robot that can physically press the power button on hardware like Mac minis.

The company also demonstrated a self-driving robot in 2023 that was cable of moving heavy server racks, so it's not the first time Meta has been exploring how to automate its growing data center footprint.

Meta continues to look at ways to automate data center tasks

According to the latest reports, some of those earlier robots have already been deployed at its Iowa and Virginia facilities, and it's the Iowa campus that looks to be one of the company's biggest testing grounds. For the past year or so, the Facebook's parent company has reportedly been testing two two-arm robots for cable-related work, under human supervision.

A separate scissor-lift-style machine with a robotic arm is also rolling around Meta's Ohio campus in a trial for jobs like reseating components.

However, the growing complexity of AI data centers also presents its own challenges for robots, making fine motor skills far more important for costly and vital components.

Naturally, worker feedback hasn't always been so positive. Besides one unnamed employee warning that robots could replace 80% of the human roles before long, many also complain that autonomous robots spend far too long charging their batteries – not working.

Still, robots are clearly still part of the company's plan. Robotics exec Eric Xu previously described a future where robots are used for incident response, environmental monitoring and preventative maintenance.

"America is in the middle of its biggest infrastructure boom since World War II, and there’s a major shortage of skilled workers to fill the roles; we need more workers, not fewer," company spokesperson Francis Brennan told Ars Technica.

Although AI is advancing at a rapid pace, the idea that it might one day replace humanity has thus far been consigned to the realm of fiction.

Rather than signalling the end of mankind, however, others have speculated that an advanced AI may one day simply supplant our position at the top of the food chain and live harmoniously with us, a species that will live to serve it.

Electric sheep

The American mathematician and computer scientist Claude Shannon was giving an interview with Omni Magazine on his work and legacy when he speculated on the role of AI and robotics in the future.

His comments came during a conversation about the limits of digital computers and the piping that backs so much of the AI technology we use today, artificial neural networks. He suggested, despite the sour mood around AI at the time, that it would be one day so advanced that our collective intelligence as a species would pale in comparison.

Far from being a doomsday scenario, though, his likening of humanity to a dog indicated a sense of companionship, mutual love and respect, and a sense of loyalty that would cement mankind's role in a world dominated by a future machine intelligence.

Roboslop

At a time when funding for AI was running dry and we encountered many bottlenecks like the lack of compute power, science fiction writers and scientists alike both started envisioning ways in which future systems could one day replace us.

Although none of those predictions have come to pass, there is a widening consensus that one day the technology will grow to surpass the level of human cognition.

This hypothetical technology, known as artificial general intelligence (AGI), may not be invented any time soon – and we may not presently have the architecture required to see progress. But scientists expect this to come within living memory.

Once we reach this point, which many refer to as the technological singularity, it's wholly unclear what the implications may be and whether AGI may eventually give rise to a cascading series of more intelligent systems that culminate in an artificial superintelligence (ASI).

• Microduck is Pollen Robotics' new programmable robot
• It looks like a cute robo-duck, and has some adorable, pre-baked interactions
• It's up for pre-order priced at $399, and it should ship in time for Christmas

If the measure of a successful robot is cuteness, Microduck could be the most successful bot in ages. Unveiled this week by Pollen Robotics (a division of AI company Hugging Face — yes, that Hugging Face), Microduck is a 25-centimeter-tall educational robot that appears to arrive with surprising mobility, balance, and personality.

Pre-orders, which kicked off today (August 27) promise a $399 (that's the introductory price), roughly 1.7lb pre-built duck-like robot complete with a mouth that doubles as a gripper, a voice —duck calls, it seems — 15 degrees of freedom (basically a measure of the number of ways all the motors let it move), and motors that let it walk, waddle, move its head and, thanks to a pair of IMUs (inertial movement unit), stay upright even when pushed around.

There's a single-eye camera paired with a Lidar system for navigating its environment and, perhaps, recognizing its owner. Pollen Robotics has yet to specify the camera's megapixels. There are also speakers for voice, and microphones for it to hear. The robot runs on a relatively low-power Rockchip RK3566 ARM chip that's backed by 1GB of RAM and 32GB of on-board storage. There's on-board Wi-Fi, Bluetooth, NFC, and a smallish 2,600mAh battery that promises roughly one hour of operation.

These are not killer specs, but the teaser videos show a highly engaged and undeniably adorable little bot. I was particularly taken with how it can recover and stand back up after a fall (or being pushed over). Microduck will arrive with a lot of pre-programmed interactions, but much of the cuteness may come via its open-source platform and your programming skills.

The Microduck kit will ship with an SDK that includes virtual training environments, along with training scripts and tools, and "a tested sim-to-real workflow." The idea, obviously, is that while you might grow to love Microducks' baked-in personality, it's what you can build with your kids that will really matter. This is an educational robot, after all.

It's almost familiar

It's also worth mentioning Microduck's obvious resemblance to another extremely cute, though entirely unattainable robot: Disney's and Star Wars' BDX Droid. Both robots are bipedal with a somewhat squat posture and cute, oversized heads, and as with the BDX-Droid, it's hard to believe the Microsoduck can stay upright. But that's where the similarities end. That semi-autonomous/remote-controlled BDX Droid is much larger, probably far more expensive, and is currently only wandering around Disney World. Plus, there are currently no plans to bring it to consumers.

The $399 Microduck is just the base package, and Pollen Robotics also plans to ship an optional charger pack for $39, a dev pack that includes spare motors, cables, and batteries for $119, and an accessory pack that includes a laser pointer (perhaps for Microduck to chase), roller skates(!) that Microduck can wear and ride, and 10 NFC tags. One wonders if those tags might be used to trigger certain Microduck actions.

Microduck is Pollen Robotics' second edu-robot, the first being the Reach Mini, an almost equally cute but more or less stationary desk companion robot. Pollen Robotics claims it sold 10,000 units.

Microduck's success will likely depend on just how many of those endearing qualities are on display out of the box, and how much we have to code ourselves. Initially, you can fake some of it, since Microduck can be remote-controlled with a gaming controller; but the real fun happens when it starts to respond in unexpected ways.

The other X factor is how many people are willing to pay almost $400 for a plastic and semi-autonomous robot duck. I know I'm intrigued, and wouldn't mind trying to program it to create the duck-bot of my dreams.

Microduck is expected to ship by Christmas, and is available in Cream, Graphite. Lavender, and Sky.

• The 2026 World Humanoid Robot Games are happening this week
• Robots are performing better than ever across 51 events
• Stopping a sprinting robot remains a major engineering challenge

The World Humanoid Robot Games are underway in Beijing, China, and records set by flesh and blood humans are tumbling: these bots have now beaten our best efforts at the 100-meter sprint and the high jump.

As the Associated Press reports, the robot record for the 100 meters now stands at 9.39 seconds (beating Usain Bolt's 9.58 seconds), while a humanoid bot has reached 2.88 meters on the high jump (above Javier Sotomayor's 2.45 meters).

This is only the second year of the Robot Games, but the progress from 2025 is noticeable — these robots were only reaching 0.95 meters on the high jump last year, for example. One engineer told the South China Morning Post that improved networking capabilities, meaning better access to cloud AI processing, is one reason for the advances in the tech.

These humanoid robots will be put to the test across 51 different disciplines over the course of the five-day event, with sports like tennis, table tennis, and kickboxing newly added for 2026. More practical trials are also included, covering tasks across industrial production, logistics, and hospitality.

Stop right there

As impressive as a lot of these robot athletes are, we're still seeing plenty of fails — such as sprinting robots smashing into safety barriers to stop. Getting these machines to slow down is actually a harder engineering challenge than you might think: we're talking about high-speed, complex movement that needs to be managed with exact precision.

There's a lot of energy that needs dissipating, a lot of rebalancing that's required, and a lot of transitioning from one state to another to be done. Researchers have managed it in simulations, but not yet in actual sprinting robots — right now the speeds and movement mechanisms are too much for the robot AIs to cope with.

It comes back to a perennial problem for robots: avoiding falling over. We take staying upright for granted, but it involves a lot of split-second calculations. When a robot is running, the problem gets exponentially more difficult.

Those watching don't seem too concerned by the abrupt stopping mechanisms. "These sports are perfectly normal for humans, but now robots can do them. I find it amazing," spectator Yang Shangzheng told the Associated Press.

• A Florida company has moved from printing furniture to manufacturing military drone boats
• Haddy built its latest unmanned boat using giant robotic 3D printers
• The company completed an earlier autonomous vessel project in nine days

Haddy, a digital manufacturing company in St. Petersburg, Florida, has produced a military drone boat using large-format robotic 3D printing.

The company announced the vessel, called the TF-179 Drone Boat, but did not name the customer behind the project.

Haddy started in 2022 building custom furniture and architectural pieces before shifting much of its focus toward defense manufacturing.

A rapid build process

The St. Petersburg facility spans roughly 30,000 square feet (about 2,800 square meters) and houses eight large-format CEAD printing systems.

Seven of those machines run on rails, extending their reach to produce unusually large single-piece components without seams.

Haddy previously used this process to print a carbon-fiber-reinforced hull for HavocAI, delivering a testable vessel from design to sea trial within nine days.

"The maritime industry is redefining how vessels are designed and manufactured," Haddy said in a company statement announcing the TF-179 vessel.

Haddy released photographs of the finished TF-179 vessel alongside its statement but offered no details about the boat's intended mission.

Demand for small, expendable maritime drones has surged across the U.S. defense industry over roughly the past year, according to the company.

A shift in Naval strategy

Haddy has also partnered with Blue Ops, the maritime drone division of Red Cat Holdings, to help double production capacity for uncrewed surface vessels.

Blue Ops sells those drone boats to U.S. and allied militaries as smaller, cheaper alternatives to traditional crewed patrol vessels.

Haddy was founded by Jay Rogers, a former Marine and Princeton graduate who previously led a similar 3D-printing venture called Local Motors.

The company opened what it describes as the world's largest 3D printing facility, based on total throughput and machine count, in April 2025.

This trend toward faster, cheaper naval hardware follows direct lessons drawn from Ukraine's ongoing war against Russia in the Black Sea.

Inexpensive, remotely piloted drone boats have repeatedly damaged or destroyed far more expensive warships belonging to Russia's Black Sea Fleet.

Even in Iran, swarms of very cheap Iranian drones have been used to track expensive aircraft, some of which were eventually brought down, including a $60 million U.S. Air Force F-15E Strike Eagle and multiple $30 million MQ-9 Reaper drones.

These outcomes have pushed American and allied defense planners to prioritize speed and volume over the years-long timelines of traditional shipbuilding.

Small, disposable drone boats can be replaced and redesigned far more quickly than a single multimillion-dollar warship ever could.

Robotic additive manufacturing fits into that calculus because Haddy can adjust a vessel's design digitally and print a new version within days.

Conventional shipbuilding for crewed naval vessels can take years to complete, a pace defense planners increasingly view as too slow for today's threats.

Battlefield requirements have shifted quickly in recent naval conflicts stretching from the Black Sea to the Middle East, adding pressure on manufacturers.

Haddy has not said whether the TF-179 design will be adapted for other military customers beyond the still-unnamed client behind this particular order.

Whether this printed approach can match the durability and long-term reliability of conventionally built naval vessels remains to be seen.

Via Defence Blog

• The Honor Robot Phone is attracting lots of attention on social media
• Commenters are calling its built-in gimbal an "innovation" and arguing that customers "are tired of seeing the same phone every year"
• Some critics argue it could be fragile or hard to clean

Earlier this month, Phones Editor Axel Metz travelled to Shenzhen, China, to be among the first to try the Honor Robot Phone.

This upcoming flagship from Chinese smartphone giant Honor features a built-in gimbal camera that can pop out of a special rear panel for enhanced video stabilization, along with high-tech features like AI-powered subject tracking and advanced shooting modes.

It's one of the most distinctive and exciting smartphones we've seen in years, and it's likely to cost less than you might think if it does reach Western markets. It officially launched in China for ¥9,999, which comes to around $1,500 / £1,100 / AU$2,100.

There is a big catch, however. Honor used to be a Huawei subsidiary and, while its devices are no longer subject to sanctions following its separation in 2020, they're still not officially sold in the US.

That fact hasn't stopped our US audience from expressing their excitement for the Robot Phone on TikTok and Facebook, where a brief hands-on video we posted last week has already reached more than a million views and accumulated thousands of likes and comments.

One top-rated comment reads: "This is why Chinese companies get banned in the US. Our tech is so behind."

"US companies can't compete," reads another. "That's why they just banned them."

Other users praised the unique concept of a gimbal built into a phone. "That's what I call innovation. People are tired of seeing the same phone every year, like the iPhone," wrote one commenter.

Another posted: "This right here is innovation light-years ahead of iPhone."

"Americans only dream of phones like these," wrote one wistful commenter.

There are also plenty of sceptics, with some users citing durability concerns and others arguing that the gimbal would be hard to keep clean. "External moving parts are never a good idea," wrote one critic.

There's currently no official word on whether the Honor Robot Phone will be available in the US, UK, or Australia, but it certainly seems like there's a lot of interest. Hopefully Honor makes an announcement soon.

In the meantime, you can read what Axel thought after two hours with the Honor Robot Phone, or our interview about the Honor and ARRI collaboration behind its camera.

The Tesla Optimus, also known as the Tesla Bot, is one of an entire catalogue of humanoid robots that manufacturers have unveiled in the last few years. Humanoid robots may look the part, and while they may be a few years away from maturation, executives like Tesla CEO Elon Musk are adamant they're a direct substitute for human labor.

Optimus Prime

Before Musk had unveiled the first prototype of the Optimus robot, he was speaking to its potential and the role it could play in the workforce of the future.

Speaking at the Wall Street Journal CEO Council Summit, Musk reportedly said that the robot could one day be good enough to replace human labor, just as machinery in the last few decades has gradually replaced humans in production lines.

His comments also suggested that rather than eliminating jobs, and thus leading to mass unemployment, the rise of humanoid robots will fill a gap in labor demand that currently exists, allowing corporations to make more and build quicker. He was adamant, in these comments and in several remarks since, that there aren't enough people and we aren't reproducing enough to maintain our current population levels.

Robotic replacement

Musk is adamant that the rise of humanoid robots will not impact the level of available jobs – but that doesn't mean his views will play out.

Many disagree, suggesting there is a material risk to current roles, including those that involve repetitive or manual tasks. But experts disagree over whether this means a net reduction in jobs, or simply a shift in the demand for human roles in the future.

Many suggest 2027 will be the year in which robots enter a serious level of production. While the Optimus robot is not currently being made, Tesla is targeting next year and beyond for mass manufacturing and distribution following several delays to date.

Despite mammoth advanced in artificial intelligence and computer vision, and a slew of humanoid robots either being piloted or hitting parts of the market, there’s still one major challenge that machines need to overcome.

It’s tactility, sensation, knowing how to hold an object. Picking up a strawberry, for example, a robot might not know how much pressure to apply, whether it’s secure, slipping, or about to be crushed. It’s something that we humans can feel through our skin, but it’s much harder to replicate on a machine level.

But it’s a work in progress, because there are companies out there like Touchlab who are working to give these skin-like characteristics to robots in a bid to make them more useful around the home, in care or in manufacturing. And it’s a lot more complex than we could ever imagine.

In Touchlab’s instance, it actually relies on quantum tunnelling technology to measure pressure, force and direction through a material that’s thinner than human skin, and with response times of less than a millisecond.

E-skin technology exists, but deployment is a challenge

Although the solution is laid out on the table, moving from pilot to production is a different ball game. Durability, reliability, affordability, adaptability… they’re all considerations that stand between theory and real-world use cases.

Touchlab is addressing some of these concerns with off-the-shelf components that already exist for established platforms, using bespoke engineering only where necessary.

It’s also seeking to reduce latency by bringing processing to the edge so that the robot only transmits the data necessary for the task, being that the e-skin can detect and collect huge amounts of micro-data.

With robots now rolling out across workplaces, care environments and, likely soon, homes, the ability to sense contact is becoming as important as the ability to see.

I speak with Touchlab’s team to understand how we can go from pilot to production in a realistic way.

• Why do robots need the sense of touch, and how is Touchlab trying to achieve this?

Computer vision allows a robot to predict, but touch is what confirms. Without tactile feedback, robotics remains trapped in a classic paradox: a machine can calculate complex equations in milliseconds, yet struggle to pick up a strawberry, handle soft medical instruments, or turn a valve without either crushing the object or dropping it. Without tactile sensing, a robot is functionally numb.

Touchlab solves this by developing biomimetic electronic skin (e-skin) that is actually thinner than human skin. Powered by quantum tunneling technology, our e-skin measures 3D forces, direction, and pressure in real-time. Crucially, it detects incipient slip—meaning the robot senses that an object is beginning to slip before it actually drops, enabling instantaneous, sub-millisecond grip adjustments just like the human nervous system.

• I tried the technology at Future Lab... I tried to control a robotic arm with haptic feedback and it felt a bit clunky with cables and restrictions. Why is it so hard to make it smoother?

Teleoperation is a classic engineering balancing act! First, there is a natural learning curve—controlling a high-degree-of-freedom robotic arm through a haptic glove isn't always intuitive on the very first attempt. It usually takes a bit of practice time and muscle memory to get the hang of the subtle force dynamics.

Second, running live demonstrations at high-traffic public events like Future Lab introduces specific hardware constraints. To ensure hundreds of visitors with completely different hand sizes can safely participate, setup requires universal-fit, heavy-duty tethered rigs and robust cabling.

Longer-term, our roadmap moves away from bulky, tethered operator suits entirely. By embedding our e-skin directly onto the robot and pairing it with onboard AI, the robot handles slip detection and dexterity autonomously—removing the need for an operator to wear restrictive physical gear.

• Right now, the focus is on your proprietary robot fingertip tech. What's the plan going forward? Do you plan to license it or sell it as a finished product? How will you deal with various robotic architectures?

Every robot platform is unique—whether it’s a humanoid deployed into hazardous industrial environments or a cobot assisting nurses in hospital wards. Touchlab is uniquely positioned to design, develop, and manufacture bespoke tactile devices tailored to specific OEM architectures.

At the same time, we offer scalable, off-the-shelf tactile components that are pre-integrated (both electronically and mechanically) with a wide variety of popular robot end-effectors. We provide volume-based discounts alongside strategic hardware loan and collaboration programs for robotics companies aligned with our vision. Over the past seven years, we have become extremely adept at rapidly building reliable integrations to suit diverse customer requirements.

• What are currently the biggest technological barriers Touchlab is encountering? What does the end game look like, and will we ever digitize our largest organ, the skin?

Our ultimate end game is for Touchlab’s tactile technology to become as ubiquitous as MEMS sensors— integrated into every commercial robot interacting with objects and humans. We envision full "finger-to-toe" tactile coverage powered by specialized edge computing and tactile orchestration software that mirrors the human somatosensory system.

Regarding barriers: the challenge isn't just technical; it's navigating western market adoption timelines for deeply transformative hardware. While digitizing human skin at scale presents manufacturing and durability hurdles, Touchlab can already deliver sensors matching or exceeding human-level spatial resolution, readout frequency, and sensitivity. The real task is packaging this capability into rugged, cost-effective formats ready for mass market deployment.

• How is Touchlab designing the user interface to reduce operator fatigue during long shifts?

We focus on shared autonomy. Rather than forcing a human operator to manually sense and adjust every micro-force vector, our e-skin handles low-level reflexes (like slip prevention) locally on the robot. The human operator directs higher-level tasks, while the robot's tactile system autonomously maintains optimal grip— dramatically reducing both physical strain and cognitive fatigue.

• High-density e-skin generates massive amounts of data. How do you handle bandwidth and network transfer without overloading systems?

High-density sensing does not automatically equal richer functional data, and often comes at the cost of hardware robustness. For our commercial products, we optimize spatial and temporal sensor density specifically for manipulation tasks, keeping data payloads lightweight.

For advanced R&D, we utilize distributed edge computing and are pioneering event-based asynchronous sampling in tactile sensing (similar to event-based vision cameras). Because our e-skin does not rely on heavy image-based data streams, it operates with extremely low latency and modest network bandwidth— making long-distance remote teleoperation surprisingly efficient.

• Because e-skin can identify objects and textures, what protocols protect tactile data security and privacy?

Tactile data on its own does not contain personally identifiable information. However, when integrated with video, audio, and LiDAR streams, multi-modal data security becomes critical.

For example, during our deployment with the Valkky medical robot in a Finnish hospital, all data transferred across hospital networks was strictly encrypted. Identifiable multi-modal data remained entirely on-premises on air-gapped storage devices and was fully anonymized before being processed for research or system analytics.

• E-skin is often seen as a high-end luxury. What is your roadmap for driving down production costs for commercial robotics?

Much like the early days of automobiles, groundbreaking technology often enters the market at a premium before scaling. While Touchlab is capable of producing full human-spec density today, deploying that level of complexity everywhere would make commercial robots cost-prohibitive.

Our strategy focuses on pragmatic engineering: delivering touch sensing tuned precisely to real-world task requirements at an accessible price point. We are actively collaborating with industry bodies to establish standardized performance metrics for tactile data, driving down manufacturing costs and making e-skin a standard baseline for all commercial robotics.

Writing about chipsets can be a bit like watching paint dry. These tiny mobile brains get more powerful every year, but 99% of smartphone users don’t get anywhere close to using their full potential. The Snapdragon 8 Gen 3, for instance, came out in 2023, but most people would be hard-pressed to notice a performance difference between that chipset and Qualcomm’s latest, the Snapdragon 8 Elite Gen 5, which powers most of today’s best Android phones.

It’s a classic case of diminishing returns — semiconductor manufacturers can keep raising the chipset performance ceiling, but if phones can’t make use of all that extra power, the tangible gains are small at best, and non-existent at worst.

It was refreshing, then, to hear Honor describe the Snapdragon 8 Elite Gen 5 chipset in its new Robot Phone as “very good” for still image processing, but “not good enough” alone for cinema-grade video capture. Did we just skyrocket through that seemingly impenetrable performance ceiling?

Well, it’s a bit more nuanced than that. As Honor’s President of Hardware Engineering, Yang Liu, explained to me in a group interview at the Robot Phone’s China launch: “The core CPU [in the Robot Phone] is the Snapdragon [8 Elite Gen 5], but Honor has developed a dedicated Image Signal Processor (ISP) — the Honor Imaging Chip H1. The processing [demands] of the Robot Phone are very high, so we must use a dedicated chipset for that, not just the Qualcomm chipset. The Qualcomm chipset offers very good performance for still image processing, but for video, it’s not good enough — for now.”

All phones have an ISP (or multiple ISPs) built into their main chipsets — indeed, the Snapdragon 8 Elite Gen 5 was the first mobile chipset to feature triple 20-bit ISPs, which, in layman’s terms, means it’s a very capable chipset for photo capture. But the best camera phones — think the Oppo Find X9 Ultra and Vivo X300 Pro — feature custom ISPs alongside those in their respective chipsets to further enhance camera performance.

That’s what Honor has done with Robot Phone, except it’s not framing its new H1 Imaging Chip as merely a complement to the Snapdragon 8 Elite Gen 5, but as a necessary addition to allow the gimbal-equipped Robot Phone to fulfil its potential as a cinema-grade video capture device.

Built in partnership with German film camera manufacturer ARRI, the H1 enables real-time on-device image processing — that’s things like white balance adjustment, tone mapping, and exposure control — and supports 10-bit ARRI LogC3 recording and ARRI Wide Gamut 3.

It’s a formidable piece of micro-engineering, and my impression is that ARRI — a 109-year-old company with exceptionally high quality standards — was only willing to partner with a burgeoning Chinese tech company on the Robot Phone because Honor could provide it with the level of processing power necessary to uphold those high standards.

“Pixels are very precious to us,” David Bermbach, ARRI’s Chief Product Officer, told me in the same interview. “[Working on the Robot Phone] was a challenge, but also an opportunity, because mass-produced devices have much more functionality on a smaller scale than special devices for one industry, where we sell a few thousand [units].

“So the task for us was to basically analyse all the different little Lego pieces on the different modules that exist in the phone and [figure out] how to connect them, how to utilize them, to basically rebuild as much of our pipeline as possible. And luckily, the processing power of the Robot Phone is extremely high.

“Also, there’s not just one camera [in the phone],” Bermbach continued, “there are several cameras. And we had to [figure out] how to get the best out of them, how to use the tools at hand, and optimize them towards [producing] a technically good picture, from our perspective, but also an artistically good picture [from a smartphone user’s perspective].”

As I wrote in my hands-on impressions of the Robot Phone, its 200MP, 1/1.28-inch gimbal camera produces dynamic, professional-grade footage — or at least footage comparable to what you can achieve with larger, dedicated gimbals like the DJI Osmo Pocket. The key difference here, of course, is that you’re getting the benefits of a gimbal and a phone in a single device, as well as ARRI’s “cinema-standard processing,” as Liu puts it.

Evidently, then, Honor’s H1 Imaging Chip is an impressive innovation, and while the Robot Phone has its problems — it gets noticeably hot after prolonged filming sessions, for instance, and isn’t anywhere near as durable as other flagship phones — I’m excited to see how far creators can push its unique video capabilities.

“This [device] is so revolutionary that we don’t even know all the use cases [yet],” Bermbach told me in our interview — he’s clearly excited too.

Humanoid robots are on the rise, with a slate of companies building several different prototypes. Whether they're designed to be on the production line or engineered to roam around your house performing various chores, different kinds of machines are on the cusp of mass production.

Setting expectations

Australian roboticist Brooks remarked on the appearance of robots in the first of his own three laws of robotics, inspired by Isaac Asimov, in a blog post that he self-published.

The co-founder of robotics company iRobot, which created the Roomba robot vacuum, Brooks explained that the anthropomorphization of humanoid robots – both by engineers and by consumers – could be counterintuitive.

While the form is necessary to gain pubic acceptance, engineers must deliver on performance and capabilities. Adopting a similar form to that of humans may imply a level of humanlike intelligence or competence in navigating the real world. Should robotics companies fail on that count, people will see straight through the ruse.

Robots of mass production

Brooks' laws focus on the competency and capabilities of robots and the need for creators to ensure that they are trained, that they're capable, and that they don't remove human agency.

Plenty of highly polished demos, recorded promotional videos, and on-stage appearances have done well to highlight the fact that these machines are not ready for prime time – despite impressive capabilities in highly controlled environments. That's evidenced by a series of 'robot fails' when they're introduced into real-world settings.

Companies, however, say they're improving the AI technology powering these machines, as well as the dexterity of appendages. They plan on ramping up production toward the end of this year, with the first machines set to be deployed in commercial settings by 2027.

First there was the phone. Then there was the iPhone. And now, according to Honor, there is the Robot Phone — a “revolutionary, category-creating device that shatters the physical boundaries of traditional mobile technology.”

Now look, I love to call out marketing department hyperbole as much as the next journalist, but notwithstanding its description as the evolutionary successor to Apple’s greatest-ever product — sell 3 billion units, then we’ll have that conversation — the Chinese tech company is right to position the Robot Phone as something wholly different to what’s come before.

No smartphone I’ve ever come across features a fully motorized, three-axis mechanical gimbal built directly into its chassis, nor such a comprehensive suite of professional-grade camera features, some of which are autonomously controlled by a physically expressive AI agent (it even has a name: YOYO). We’re in full The Hitchhiker’s Guide to the Galaxy territory here, and while the Robot Phone may never become as mainstream as the iPhone, it most definitely is as revolutionary as Honor wants us to believe.

First teased in October 2025, the Robot Phone appeared in prototype form at MWC 2026, before finding its way to the Cannes Film Festival and, more recently, the World Cup final (Spain’s Eric Garcia was spotted filming with the device). Now, it’s officially launched in China for ¥9,999 (around $1,500 / £1,100 / AU$2,100), and I was among the first group of journalists permitted to hold, capture, and critique the Robot Phone at its extravagant Shenzhen showcase.

Gimme, gimme, gimme a gimbal

Let’s start on that gimbal, which marks the Robot Phone out as something entirely different from any of the best phones money can buy in 2026. It really is as if Honor has strapped a DJI Osmo Pocket to its flagship Magic 8 Pro, except the gimbal in question can tuck itself away into the phone’s rear panel when not in use, almost like an S Pen slipping discreetly into a Samsung Galaxy S26 Ultra.

Why would you want a gimbal attached to your phone? For several reasons. There’s the improved video stabilization, which results in steadier footage; the manual 360-degree control of the gimbal camera itself; the AI-powered subject tracking; or even the many automatic shooting modes like AI SpinShot (a 90- or 180-degree camera roll) and Tilt Locked (which locks the phone’s gimbal at your chosen angle).

This is no knock-off gimbal, either. Honor says the Robot Phone’s mechanical gimbal system — which features more than 100 individual components — is some 65% smaller than mainstream gimbal systems like the Osmo Pocket, and its structure is 200% stronger as a result. The gimbal motor itself weighs just 2.6g and facilitates movement speeds of up to 360 degrees per second, while a dedicated Flip Motor delivers a torque density of 120 Nm/L.

Full disclosure: I did not know what torque density meant before seeing the Robot Phone for myself, but I now know it’s a measure of how much twisting force a motor can produce relative to its size. A figure of 120 N·m/L is, apparently, very good — it means the Robot Phone’s motor is powerful enough to turn the gimbal with speed and precision, despite being small enough to fit snugly into the phone’s rear camera module.

And good news! This mechanical gimbal system works very well. I was skeptical about how effective a phone-mounted gimbal could possibly be, but after playing around with the Robot Phone’s various shooting modes, it produces dynamic, professional-grade footage — or at least footage that’s comparable to what you can achieve with larger, dedicated gimbals like the DJI Osmo Pocket. The key distinction, of course, is that the Robot Phone doesn’t ask you to walk around with (or purchase) a separate gimbal.

The gimbal camera is a 200MP 1/1.28-inch sensor with an f/1.6 aperture, supporting up to 4K video with special recording formats like LogC3 and ARRI Cinema. The latter comes as a result of ARRI’s partnership with Honor on the Robot Phone; the German brand — known for producing motion picture film cameras — lends its expertise to help bring Hollywood-grade color science to Honor’s smartphone pipeline.

Not to get too technical, but Honor says the Robot Phone performs noise reduction earlier in the RAW domain, processes data in 14-bit 4:4:4, and outputs 10-bit LogC3 and 4:2:2 video, preserving more usable image information for post-production. ARRI LogC3, meanwhile, preserves more highlights and shadows, while ARRI Wide Gamut 3 provides a broader color spectrum.

If that all sounds like a foreign language to you, that’s because the Robot Phone is absolutely intended for video creators, rather than regular folk; its video stabilization capabilities are certainly useful for everyday shooting, but for me, they’re not leagues ahead of what you’ll get from software-based stabilization features in other phones, like Apple's Action Mode or Samsung's Super Steady with Horizontal Lock. The real juicy benefits are those post-production features, which were hard to capture and quantify in my short hands-on session, but were certainly present.

Say ‘hello’ (again) to YOYO

I did, however, get a taste of the Robot Phone’s YOYO agent. This agentic assistant initially debuted on the Magic 8 Pro in China as a tool for carrying out cross-app smartphone tasks on your behalf — it’s not dissimilar to Google’s Gemini Intelligence feature in that way — but on the Robot Phone, YOYO uses the existence of a gimbal to express itself physically. At launch, YOYO can nod, shake its (gimbal) head, dance to music, and perform other motorized oddities, which are complemented by over 100 on-screen facial expressions.

The desired effect is that YOYO feels like a physical companion in your pocket, unlike any other smartphone-based assistant on the market. The reality is that these physical gestures and on-screen expressions are the definition of gimmickry, though the ability to give YOYO gimbal-related commands using natural language — “turn left 45 degrees” or “turn around for five seconds,” for instance — is genuinely useful, and worked well.

Outside of its gimbal and YOYO agent, the Robot Phone is every bit a 2026 flagship. It’s got a 6.3-inch display; it measures just 9.59mm thick (for context, that’s less than 1mm thicker than an iPhone 17 Pro Max); there’s a Snapdragon 8 Elite Gen 5 chipset, a sizeable 7,060mAh battery, and two additional flagship-grade cameras (a 200MP telephoto and 50MP ultra-wide).

In the hand, the Robot Phone doesn’t feel too dissimilar to existing 6.3-inch phones on the market, which is quite the feat given that there’s a full-on gimbal on offer here. That said, I’m not sure how well that gimbal is going to hold up to real-world use — I felt, for instance, like I could easily (or accidentally) snap off the plastic shield covering it. And that’s to say nothing of its vulnerability to dust and debris when fully extended.

But none of this really matters. The Robot Phone is not meant to be treated like a regular phone, because it is not a regular phone. Heck, almost nobody reading this can even buy it. It’s a ridiculous, remarkable product that should be thought of more like the Samsung TriFold — i.e., an example of what’s technologically possible — than as a genuine disruptor to an industry that foldable phones have barely managed to dent.

The Robot Phone is essentially Honor saying, “Look at what we’ve been able to do with technology,” and I don’t think there’s anything wrong with that. To have brought a fully functioning robotic phone to market is no mean feat, and it’s exciting to see and touch a device that’s so drastically different from the phones we’ve become used to.

In many ways, the Robot Phone is like one of those outlandish gadgets from science fiction movies. And using it made me feel like a kid again.

• Camera-only systems can only “approximately match” a human driver

• Multi-sensing setup offers more than just a failsafe
• Waymo co-CEO backs the use of Lidar and radar over camera-only systems

Waymo co-CEO Dmitri Dolgov has spoken out about the technology required to run a fully autonomous robotaxi service, addressing the subject of camera-only systems.

While he didn’t mention Tesla directly in his in-depth talk at Y Combinator’s Startup School, the presentation directly addressed vision-only camera setups, like those Elon Musk and Tesla have been pursuing.

“Humans, of course, can drive with just eyes, so there’s that proof of existence,” he said, according to Electrek.

“If the goal were to just approximately match human performance or to build an assist product, that’s a very reasonable way to go.”

But “weak sensing” inevitably leads to a safety curve that flattens out “way too early,” he added, highlighting the many cases where a camera system can come unstuck.

Low light or poor visibility, dust storms and objects physically blocking the lens can bring a camera-based system to a grinding halt in seconds.

Dolgov then gave an example of how Waymo’s combination of Lidar, radar and high-definition cameras allowed one stricken vehicle to safely navigate itself back to a depot when a tree branch became caught in its wipers and obscured the camera lens.

While Dolgov doesn’t completely dismiss camera-based systems, stating that they can easily support a driver with highly autonomous assistance systems, he admits they will struggle to replace a human entirely.

The issue is that fully autonomous robotaxis need to be as close to perfect as it’s possible to be.

A mounting record of accidents will likely put off customers for life and kill the technology before it has time to gain a foothold.

Many take the multi-modal path

Having ridden in numerous autonomous vehicles over the years and listened to those that have built them, I am fully in Dmitri Dolgov's camp.

While a camera-based system might help keep costs down, fitting autonomous vehicles with Lidar, radar, and other sensors feels like a natural and reasonable step to make them as safe as possible.

A visit to Motional’s HQ in Las Vegas earlier this year illustrated the need for multiple sensing technologies when you completely remove the driver from the picture.

Its robotaxis, which are now bespoke versions of the Hyundai Ioniq 5 (complete with radar and Lidar), have driven over two million autonomous miles with zero at-fault incidents.

Like Waymo, it too believes a multimodal approach that fuses various technologies is the safest way forward.

Laura Major, the company’s CEO, said at the time that rival startups were able to get to a “90% performance” very quickly, often relying solely on high-definition cameras.

“These offer really smooth rides in most situations,” Major remarked. “The problem is when you’re going to remove a driver and have true driverless operation, that’s not enough. You can’t perform well in 90% of the cases,” she added.

Admittedly, I haven't ridden in one of Tesla's driverless Robotaxis, but the numbers don't lie.

According to Electrek, Tesla's own data confirms a crash rate three times worse than humans even with a monitor at the wheel, while Waymo's numbers suggest it is 17 times safer after covering some 220 million rider-only miles.

Tesla has accumulated just 380,000 driverless miles over the past year since launching its robotaxi service.

• Japan ramps up drone production to reduce dependence on Chinese manufacturers
• Defense startups are shifting from civilian projects into military drone manufacturing rapidly
• Japan dramatically increases funding for interceptor drones and autonomous defense systems

Japan's drive toward large-scale drone manufacturing has gained fresh urgency, as a growing cluster of startups, including Eams Robotics and Terra Drone, are now shifting from civilian projects toward defense-oriented drone production.

A report from Nikkei Asia said Japanese Prime Minister Sanae Takaichi seeks to reduce the country's reliance on Chinese suppliers.

Chinese firms currently hold about 90% of the global industrial drone market, a dominance Tokyo now views as a strategic vulnerability worth addressing.

Startups pivot from civilian work

Eams Robotics began in 2016 as an aerial photography company serving television broadcasters before moving into drone manufacturing two years later.

The company, based in Minamisoma, has since worked with Sagawa Express, the University of Tokyo, and NTT DoCoMo on autonomous flight systems.

In December 2025, the company flew a fully autonomous drone across an active railway line without a ground observer present.

It now reportedly aims to partner with the American defense firm Anduril, mirroring a wider shift taking place across the industry.

Terra Drone, the more defense-focused of the two firms, is launching mass production of Japan-made batteries for defense and industrial customers.

Batteries account for more than a quarter of a drone's cost, and Japan currently lacks a domestic supplier able to meet defense-grade demands.

Budget and battlefield pressure build

Japan's budget for unmanned defense systems jumped this year to 277 billion yen (~$1.72 billion), up from 111 billion yen (v$705 million) in 2025.

Its total defense spending is also set to rise to 3.5% of GDP, a shift expected to channel further investment into unmanned systems.

The ruling Liberal Democratic Party approved a plan on June 9 to produce roughly 80,000 drones annually by 2030, though funding details remain sketchy.

Japan's Acquisition, Technology and Logistics Agency (ATLA), the defense ministry's procurement arm, has scheduled demonstrations next month, aiming to fast-track acquisition and deployment of interceptor drones near radar sites and bases.

These systems are meant to counter long-range suicide drones such as the Shahed-136, which have flown as far as 1,100 miles in the Middle East.

The push for domestic production comes as reports surface that Tomahawk missile deliveries to Japan could face delays, after the 38-day U.S. campaign against Iran strained the U.S stockpiles.

Defense officials in both Washington and Tokyo have denied any disruption, though Japan's defense minister said domestic missile development will continue.

Unconfirmed reports suggest Anduril has discussed converting Nissan's Oppama plant into a drone production site, amid concerns over a possible Taiwan Strait conflict.

This fits into Japan's wider five-year military buildup, set to wrap up in 2027, which also covers missile networks and unmanned ground and maritime systems.

Terra Drone CEO Toru Tokushige described interest from Japan's defense ministry as "very much positive," citing lessons drawn from Ukraine's battlefield experience.

However, whether that optimism will eventually translate into real, battle-tested capability against long-range drone threats remains genuinely uncertain.

• Disney Worldbuilders lands on Disney+ on August 20, 2026
• Leslie Iwerks' new documentary goes inside Disney Imagineering
• Expect behind-the-scenes stories from Galaxy's Edge, Pandora, Pixar, and more

If you're a fan of the many franchises under the Disney umbrella — Marvel, Star Wars, Pixar, or Disney Animation — or simply love visiting Disney Parks around the globe, the company has another documentary arriving on Disney+ later this month.

I say "another" because earlier in 2026 we got a fresh look at how Disneyland was created with Disneyland Handcrafted. Now, the same director, Leslie Iwerks, has sat down with many of the creative minds behind Disney's biggest franchises, including Kevin Feige, Dave Filoni, Pete Docter, Jon Favreau, Jared Bush, Jennifer Lee, James Cameron, new Disney CEO Josh D'Amaro, and former CEO Bob Iger.

It's an impressive lineup of people who helped create many of these iconic worlds and, just as importantly, bring them from the screen or page into the physical environments we can actually experience.

From Pixar Pier and Toy Story Land to Star Wars: Galaxy's Edge, Zootopia, Pandora, and countless other destinations, Disney Worldbuilders promises to show how these lands came to life through the people who imagined and built them. It also looks set to tease upcoming expansions and entirely new experiences.

As you can see in the trailer, Worldbuilders explores both these iconic lands and the creative process behind them through interviews with the people who made them possible. Jon Favreau, who most recently directed The Mandalorian & Grogu, says, "We can push the limits of technology to pursue something mind-blowing," and that feels like an apt description of Disney Parks today.

Just look at the recent update to Pirates of the Caribbean at Disneyland with its new projection-based Audio-Animatronic, or the overhaul of Millennium Falcon: Smugglers Run, which now feels even more like a video game thanks to Unreal Engine-powered updates. That same drive to push technology forward can also be seen in the free-roaming BDX Droids in Galaxy's Edge, the walking, talking Olaf robot, and even experiences aboard Disney Cruise Line ships.

There's a good chance Worldbuilders explores many of those innovations. While it will almost certainly lean more heavily on interviews than Handcrafted, Leslie Iwerks' involvement gives me confidence it'll have the same warmth, care, and strong narrative that made that documentary so enjoyable. That's especially true if you're a fan of Disney Parks, Disney's many franchises, or simply enjoy documentaries about how ambitious creative projects come together.

Beyond the technology, it also appears the series will follow the journey from early concept art to construction and finally opening day, while showing how these lands continue to evolve through new characters, refreshed attractions, and entirely new destinations. I'm especially curious whether the Pixar section will include a look at the upcoming Monstropolis land at Disney's Hollywood Studios, as well as The Magic of Disney Animation, which opens there next month.

Disney Worldbuilders premieres exclusively on Disney+ on August 20, 2026. Unlike Disneyland Handcrafted, it won't also be available to watch for free on YouTube. If you're looking for something to watch before then, I'd also recommend We Call It Imagineering, Behind the Attraction, or Disneyland Handcrafted for a behind-the-scenes look at how Disney's magic is made.

• China unveils rugged AS2-W robotic platform with advanced onboard artificial intelligence
• Robot wolf climbs steep slopes while carrying heavy operational payloads
• AS2-W combines wheeled mobility with advanced terrain handling capabilities

Chinese robotics company Unitree has unveiled its new AS2-W, an uncrewed ground vehicle on its official website.

The four-wheeled vehicle weighs 25 kg and can carry a stationary payload of up to 150 kg while sitting still.

It travels roughly 30 km on a single charge without cargo, or about 25 km carrying up to 16 kg.

Sensors and mobility built for rough terrain

According to Unitree, the AS2-W can climb obstacles reaching 80 cm high while traversing slopes of up to 45° without manual intervention.

The robot reaches speeds exceeding 6 m/s using industrial-grade joint modules delivering approximately 95 N·m of peak joint torque.

Unitree also says the platform uses a 648 Wh, 15,000 mAh battery capable of providing more than three hours of unloaded operation.

Its equipment includes industrial-grade 64- to 128-line LiDAR, an HD camera, Wi-Fi 6, Bluetooth 5.2, GPS, 4G connectivity, microphones, speakers, and front lighting.

The AS2-W supports an IP54 dust and water resistance rating and is designed to operate in temperatures ranging from -20°C to 55°C.

Unitree also provides developer software tools, application programming interfaces, and support for artificial intelligence models enabling autonomous interaction and decision-making.

A demonstration video released by the company shows the robot navigating rocky slopes without losing its footing throughout the test.

The same footage shows the machine maintaining full mobility even after falling from elevated platforms during separate testing runs later.

Unitree describes the AS2-W as running a high-performance AI module capable of 150 TOPS, or trillion operations per second, onboard.

The company has not specified whether that processor is actually manufactured by Nvidia, despite widespread reports describing it that way.

Its ISS 3.0 autonomous navigation system is designed to provide centimetre-level positioning accuracy during movement through complex operating environments.

Military interest follows the armed version sighting

The People's Liberation Army has already shown interest in the AS2-W following its appearance during the Chinese-Mongolian Steppe Partner 2026 joint military exercise.

Chinese military reports indicated similar robotic platforms performed reconnaissance and fire support tasks during the exercise alongside conventional military units.

The public unveiling followed those exercises, although Unitree's commercial announcement focused primarily on the platform's technical characteristics rather than operational deployment.

The vehicle's onboard artificial intelligence enables object processing directly on the platform, potentially reducing delays associated with transmitting information to remote operators.

Its combination of autonomous navigation, onboard computing, and off-road mobility reflects broader interest in unmanned ground vehicles across several defence programmes worldwide.

However, the company has not disclosed the full operational limitations, endurance under sustained military use, or performance while carrying heavier payloads.

Unitree's YouTube video of this robot comes with the disclaimer, "Please, everyone be sure to use the robot in a friendly and safe manner."

I watched two robot videos at the opposite ends of the Cantril Scale. You know, the ladder that measures human happiness in steps but that I am now applying to humanoid accomplishment?

At the bottom step and living its worst possible artificial life, we have this Qualcomm SoC-powered robot that, after delivering the new Qualcomm Dragonwing IQ10 robotics reference platform on a tray to a presenter, spectacularly collapsed lifelessly at the guy's feet. The robot raised one hand as if in defeat, and seconds later, handlers rushed out to throw a sheet over its lifeless mechanical body.

At the top of the ladder and full of artificial life satisfaction, we have the Apollo Robot from Apptronik. It's running DeepMind's Gemini Robotics 2, a new artificial intelligence layer that lets it autonomously accomplish a wide variety of human-like tasks, which it demonstrates in a new DeepMind video.

Watching the two videos side-by-side, I was struck by both the leaps in humanoid robotics we're making almost daily thanks to new foundation models and the ability to simulate robotic training at scale, and how many humanoid robots remain wildly disappointing.

As I told someone this week, I grow tired of all the companies promising robots in the home doing our bidding (and possibly doing all of our jobs) this year, let alone within the next five years. I've been covering robotics for a quarter of a century; I get the technology, I've watched the trajectory, and even with the most recent rapid acceleration, the majority of humanoid robots fall into the Qualcomm-chip-powered and not the Apollo and DeepMind camp.

It's not clear why the Qualcomm-chip-powered robot collapsed. It could've been an onboard error, which would not be a great way to sell Qualcomm's latest robotics SoC, or maybe it was simply the battery on the human-sized bot. Perhaps someone forgot to charge it up before the presentation, and when the battery was spent, all those electric servos keeping it upright and balanced gave out.

It's not like it was some exciting demonstration of the Qualcomm technology in the seconds before that, when the robot very slowly and carefully walked out on stage looking like someone's 1960s projection of future robot capabilities.

I suspect it would have done better if it were built based on DeepMind's latest breakthrough.

Gemini Robotics 2 is interesting because instead of disparate sensors making choices about movement and touch while someone teleoperates or preprograms the major movements and tasks, this system promises wholly centralized, whole-body control.

According to DeepMind's website on the new technology, Gemini Robotics 2 "brings whole-body intelligence to humanoids, enables advanced dexterity, and can coordinate multiple robots to work together in shared spaces."

The biggest stumbling block (no offense to Qualcomm's robotics technology) is the complexity of our world. Our homes, offices, streets, and overall environments are not carefully planned out to help us avoid mistakes. Instead, they're open environments where we constantly learn and adapt. Basically, we figure out how to live, work, and interact. At a fundamental level, that's what DeepMind hopes Gemini Robotics 2 produces.

They see Gemini Robotics 2 as "the brain that controls the whole body of the humanoid" and, through the use of sensors, vision, and subroutines, adapts and learns how to navigate. In one example they showed, an Apollo robot uses its hands to cinch and tie up a garbage bag. It's something humans do with ease, but it represented a breakthrough for these robots, because they had not been trained on the task.

Meanwhile, that Qualcomm chip-powered robot is lying on the floor after a short walk.

Gemini Robotics 2 is still in the research phase, but it points to more useful and potentially safer robots that can live and work among us and other robots without the constant need for training and oversight.

That's the true state of humanoid robots, and not the promises made by other competing humanoid robot companies like Tesla, Neo, and Figure, though their development may accelerate if they adopt systems like Gemini Robotics 2.

Update 7-31-2026 5:20PM ET: After we posted the story, Qualcomm sent us a note clarifying that the robot is not a Qualcomm humanoid, though it's running the chip maker's SoC, and added this statement explaining what happened with the robot:

“Live demos involve risk. At Computex in Taipei, a brief communication glitch triggered a fault in the prototype humanoid and initiated a controlled shutdown. The robot performed its designed ‘safe-collapse’ sequence, lowering itself to its knees to protect the environment around it. The safety systems worked as intended. The incident also provided valuable real-world data that is helping to further strengthen system reliability and accelerate development of safer, more resilient robotics."

Robot lawn mowers are emerging as one of the most exciting smart-home categories. The concept has been around for a while, but in the past two years we've seen advances in technology that are making them more helpful and more user-friendly than ever. Finally, the barriers to entry are coming down, and lawnbots are starting to look like a feasible option for even the least tech-minded amongst us.

To find out more about what's going on in this growing space, I spoke to KK Yin, marketing director at major lawnbot brand Mammotion. We chatted about the most impactful shifts in lawnbot technology, the hottest trends, and the hurdles the industry still needs to overcome.

Cutting-edge edge cutting

For KK, the big feature that hasn't quite been nailed yet is edge-cutting. Even if you have a great robot mower that does a stellar job trimming your grass, its usefulness is limited if you then have to get out a trimmer to take care of the edges. "Edge trimming remains one of the biggest challenges, especially along walls," says KK. "The key is balancing cutting performance with safety."

If you're trimming along the edge of a flat path, it's relatively easy — certain mowers are designed to straddle the path edge, with the blades underneath trimming the grass at the boundary line. However, that doesn't work when trying to trim next to a wall. "The lawnmower can get as close as possible, but the cutting disc being set back from the mower's edge to maintain a safe clearance prevents the mower from cutting close enough to walls," KK continues.

Lots of brands have been working on a solution, and it typically takes the form of an edge-trimming module that attaches to the side of the bot, with smaller blades in a protective casing.

On the Luba Mini is an edge-cutting disc that can trim as close as 5cm (1.97 inches) from a wall — a distance KK admits might not be close enough. "Five centimeters is not very far from the wall, but still some users want us to make it perfect so it can cut maybe even down to one centimeter," KK says.

"We're exploring more flexible solutions," he continues. "For example, when mowing alongside a wall, the cutting disc could potentially extend slightly out of the mower's body for a cleaner edge cut. In open areas or during ride-on-the-edge mowing, it would retract to maintain a safe clearance beneath the mower. So that is the direction and also the challenge in the future."

Tackling tricky terrain

Another major lawnbot trend that's been emerging over the past year or so is all-wheel drive (AWD). This helps lawnbots tackle tricky or uneven terrain, as having all the wheels individually powered means that if one or two end up off the ground, or struggling to get traction, the others can help the robot escape without requiring manual intervention.

AWD can also help the lawnbot retain its grip when scaling especially steep lawns. KK notes that in the European market, and Germany especially, lots of houses stand on slopes, where this boosted climbing ability comes in handy.

All-wheel drive is usually indicated by an 'AWD' on the product's name, and it appears in Mammotion's advanced Luba line — you can see the LUBA 3 AWD scaling an astroturf mountain in the video above.

The brand has also been looking at upgrading its bots' climbing ability and mowing speed, and is keen to figure out what the next advancement might look like. "We can see some other companies are exploring other things like independent steering," says KK. "Maybe that is a good solution for difficult terrain. So we are also thinking about how to combine this technology."

Better tech in smaller bots

Initially, Mammotion kicked off its lawnbot journey with its premium Luba series. Later on, the Yuka Mini arrived — a smaller, more affordable bot that's designed for yards 500 to 800 square meters in size. It proved to be the big winner in Europe during the recent Amazon Prime Day sales.

"I think when we started, our premium product line helped us build trust in our technology — things like the all-wheel drive and the positioning system and the efficiency. So customers believed our brand could bring some of that technology to an entry-level lawnmower," muses KK.

The Yuka Mini is a good fit for European yards, which on average are 300 to 500 square meters in size. "Many robot lawnmower brands spend a lot and have a different product strategy for European regions, because many of the users have already accepted this innovative product for their lawn care. For example, they know about the positioning systems and how all-wheel drive compares with rear-wheel drive. They want their robot lawnmower to do more things — not only mowing, but also edge trimming."

In the US, where the average lawn size is much larger than in Europe, the focus is a little different. "I think most US users still prefer a ride-on or traditional lawnmower. I think they're still looking for an efficient robot mower that can cover their large lawn," says KK. "The US is at a different stage, and what customers require is different.

No-fuss installation

Navigation tech is perhaps the area that has moved on most quickly and has had the most impact on lawnbot usability. It was only a few years ago that boundary wires were required; now they're all but obsolete.

Initially, the technology to take over was RTK. This is satellite-based positioning, and it requires a separate RTK receiver to improve the accuracy of the satellite data. Not everyone wanted to set up a receiver in their yard, leading to a shift to centralized receivers that could serve massive areas at once. Mammotion calls this NetRTK.

Either way, the system relies on satellites, which means it has some limitations in yards where there is coverage from trees or buildings. Most recently, we've started to see LiDAR being used. This is the navigation approach used in robot vacuums, and just like with RTK, there are pros and cons when used in lawnbots. LiDAR works by bouncing light off objects to build up a picture of its surroundings, which is great when there are objects around but less useful in wide, open spaces with nothing to bounce off.

"You know, like we always say, maybe laziness is the best way to improve the technology!" says KK. "For our users, if they're happy with the intelligence of the robot lawnmower, then they want an easier installation, or even no installation at all. So that's why they really like our NetRTK, because they don't need to install an antenna. Or some users will choose LiDAR, which doesn't need an antenna at all."

The real shift is less to do with using one specific navigation approach over another, and more about moving to a situation where multiple navigation technologies can work alongside one another.

"We want to think about the final result of the positioning system," says KK. "Many users don't want to compare which positioning method is good. They just use it; that's enough. They don't need to know more about it. So that's why we came up with Tri-Fusion, which combines the advantages of RTK and LiDAR and even AI vision. We believe this can cover all environments for a residential user."

A few robot mower companies have started experimenting with similar concepts, where the bot can swap between different approaches depending on the situation. If satellite signal drops out under some trees, LiDAR can swoop in to save the day and get the lawnbot back on track, for instance. Cameras (often backed by AI) form the third pillar, enabling the bot to recognize objects and understand if it should mow close to them or steer clear.

There are problems still to solve in the world of robot mowers, but let's not forget that just a few short years ago, we were still painstakingly laying boundary wires and watching the lawnbot bounce haphazardly between them. Now, the issue is whether a bot can trim quite close enough to the edge of a wall, and if it's smart enough to tell a soccer ball from a lounger. We've come a long way.

Brilliant bots

We've tested plenty of impressive lawnbots at TechRadar. Here's a taster of some of our favorites — hit the View details button for a link to our full review for each.

In an era increasingly dominated by AI and automation, it’s still incredible just how much construction work remains manual. The contrast is most stark in data centers – facilities built to power cutting-edge technology, but whose delivery is generally still slowed down by major fragmentation and time-consuming manual tasks.

One of the industry’s biggest challenges seems to be capturing and understanding what’s actually happening on a project. Project managers can spend countless hours walking sites, checking completed work and coordinating contractors operating to different schedules. When documentation falls behind, small issues can go unnoticed and eventually develop into costly delays.

While the industry is finally starting to get to grips with this technology, construction is presenting own on challenges. Autonomy works best within fixed parameters and with a limited number of variables, but live sites offer the opposite – changing plans, moving materials, new structures being built and multiple trades working alongside each other.

Construction is where automation fails – is now the turning point?

These are the exact conditions that have made automation ineffective in construction. But automation isn’t impractical or impossible, it just means vendors will need to focus on the tasks where machines can deliver the best results.

Progress capturing, side documentation and routine inspections are some of the areas where automation could work best, and better still, companies like OpenSpace argue much of this work can actually be done outside of normal operating hours to both reduce disruption, and to reduce exposure to a live and dynamic environment.

Data centers could be a good proving ground for this, with large floor plans, repeatable layouts and intense schedule pressures. With up-to-date data from overnight checks, for example, leaders could assess progress and identify areas that need extra attention.

But collecting that data is only the first step, because the AI running behind the scenes need to be able to interpret site imagery to cross-reference it with site plans, drawings, schedules and other information held across other systems.

In this Q&A, OpenSpace CEO Jeevan Kalanithi explains why construction automation is starting to gain momentum, why ‘good enough’ may indeed be good enough without needing immediate perfection, and how automated monitoring could actually help soften the blow of ongoing labor shortages.

• Construction has traditionally been one of the least automated industries. Why is that beginning to change now?

I actually think the idea that construction is slow to adopt technology is a bit of a misconception. Builders adopt tools that genuinely make their jobs easier – they've just been waiting for technology that understands how they actually work.

Construction is fundamentally different from industries like manufacturing. Every project is unique. Jobsites change every day. Teams are working in environments that are constantly evolving, with dozens of trades operating simultaneously. That's a much harder environment to automate than a factory or warehouse where conditions are highly controlled.

For a long time, most construction software focused on documents, schedules and reports because that's what computers could understand. But construction isn't really a document problem – it's a physical-world problem. The most important information lives on the jobsite itself: what's been built, what's changed, where work is progressing and where risks are emerging.

What's changing now is that AI is becoming capable of understanding the physical world. Instead of asking people to manually document what's happening, AI can interpret images and other real-world data to understand the state of a project. That makes robotics and automation much more practical because they fit naturally into how builders already work instead of forcing them to adopt entirely new processes.

• Robotics in construction has been discussed for years but has seen relatively limited adoption. What factors are contributing to the increased interest in autonomous and semi-autonomous systems on jobsites today?

The conversation has become much more practical.

Ten years ago, people asked whether robots would replace construction workers or build entire buildings autonomously. Today the question is much simpler: How can robots help experienced builders work more efficiently? That's an important shift because construction has always adopted tools that solve real problems.

The biggest opportunities today are around repetitive, time-consuming tasks like documenting progress, capturing site conditions or performing routine inspections. Those activities are incredibly valuable, but they're not necessarily the best use of a superintendent's or project engineer's time.

I think of robotics a bit like the introduction of nail guns. Nail guns didn't replace carpenters – they helped carpenters work faster and more consistently. Robotics is following a similar path. The goal isn't to automate construction. It's to automate specific tasks that allow skilled professionals to spend more time coordinating work, solving problems and making decisions.

We're already seeing that with the robotics companies that we integrate with, each approaching different aspects of autonomy. Rather than trying to solve every problem, they're proving that robots can reliably perform specific tasks that create immediate value on today's jobsites.

• Data centers and other large-scale projects have been cited as promising early use cases for construction robotics. What characteristics make those environments more suitable for autonomous technologies?

Data centers are actually a great example of where robotics and AI can demonstrate value today because they combine three characteristics that work well for autonomous systems.

First, they're relatively structured environments. Compared to a renovation project or an occupied hospital, data centers typically have large floor plates, repeatable layouts and fewer unexpected obstacles, making them easier for robots to navigate.

Second, they're incredibly schedule-sensitive. AI infrastructure is expanding at an unprecedented pace, and every day matters. Owners and contractors need continuous visibility into progress because hundreds of activities are happening simultaneously.

Finally, they're highly repetitive. The same systems and construction sequences occur over and over, which allows robotics to operate more consistently and makes it easier to measure progress over time.

Autonomous data capture is particularly valuable in these environments because it creates a consistent visual record without disrupting work during the day. That gives project teams objective information about what's actually happening on site, helping them identify issues earlier, coordinate more effectively and keep stakeholders aligned.

• Construction sites are highly dynamic environments. What are some of the biggest technical and operational challenges robots face when deployed on active projects?

Construction sites are probably one of the hardest environments you could ask an autonomous system to operate in.

Unlike a warehouse, where everything is designed to be predictable, construction sites change constantly. Materials move. Equipment gets relocated. Walls appear. Doors that were open yesterday might be closed today. You also have dozens or sometimes hundreds of people working alongside autonomous systems, so safety has to remain the highest priority.

There are practical challenges as well. Connectivity isn't always reliable. GPS often doesn't work indoors. Navigation has to account for changing layouts and temporary obstacles that don't exist in more controlled environments.

That's why I think we'll continue seeing supervised autonomy for quite some time. Humans are still remarkably good at adapting to unexpected situations, and construction has plenty of them.

One lesson we've learned is that "good enough" often beats "technically perfect." Builders don't need the most sophisticated robot, they need one that's reliable enough to show up every day, operate safely and consistently, and fit into the way projects already run.

• As robotics adoption grows, what role does visual intelligence play in helping machines understand and navigate the physical world?

Whether imagery comes from a person carrying a 360° camera, a robot, a drone or another autonomous system, collecting the data is really only the first step. The real challenge is understanding what that data means.

AI has become remarkably good at understanding language, but if it's going to operate in the real-world economy, it also needs to understand physical places, physical assets and physical work.

Construction has always had information about what was supposed to happen – drawings, BIM models, schedules and specifications. What's historically been much harder is understanding what actually happened in the field.

That's where visual intelligence comes in. AI can compare what's been built against what was intended to be built, measure progress over time, identify potential issues and surface insights that help project teams make better decisions.

Our philosophy is simple: we don't really care how the data gets collected. Whether it comes from a phone, a 360° camera, a drone, a laser scanner or a robotics platform from one of our robotics partners, our goal is to bring that information together into a common understanding of the jobsite.

Today, our customers have captured imagery across more than 70 billion square feet of construction. That scale creates an opportunity not only to help builders today, but also to help train and validate the next generation of AI systems that need to understand the physical world.

• There's often concern that robotics could replace workers. How are builders currently thinking about the relationship between automation and the existing construction workforce?

I think that's a very understandable concern, but it doesn't reflect what we're seeing on jobsites.

The builders we work with aren't trying to replace experienced people. If anything, they're trying to figure out how to help those people do more.

Construction has faced labor shortages for decades, and demand continues to outpace the available workforce. There simply aren't enough skilled people entering the trades to meet the amount of work that needs to get done. That's especially true as we build more data centers, manufacturing facilities and infrastructure.

Robotics and AI help address that challenge by taking on repetitive work like routine documentation, progress capture or inspections, allowing experienced professionals to spend more time coordinating work, solving problems and applying their expertise.

The best technology doesn't replace people – it amplifies what people are already good at. That's how I think construction will continue to adopt AI-powered tools and robotics. They become another set of tools that help experienced teams make better decisions, work more efficiently and get more done with the resources they already have.

• The concept of "human-in-the-loop" robotics is gaining attention across industries. How does that approach apply within construction environments?

I think human-in-the-loop is going to be the dominant model for construction for quite some time.

Construction is simply too dynamic to expect fully autonomous systems to handle every situation. Something unexpected happens every day – a blocked corridor, a relocated piece of equipment, a new safety barrier, an area that's suddenly inaccessible. Humans are still exceptionally good at recognizing those situations and adapting in real time.

What we're seeing today is a very practical division of responsibilities. Robots handle routine, repetitive tasks consistently, while people provide judgment, context and intervention whenever it's needed.

Some of our robotics partners are already operating this way, where autonomous systems perform most of the work but remain remotely supervised so a person can step in if something unexpected occurs.

I don't think that's a compromise. I think it's actually the right model. The goal isn't autonomy for its own sake. The goal is giving project teams better information while maintaining the flexibility and judgment that complex construction projects require.

• Some contractors are experimenting with autonomous data capture during off-hours, such as overnight or before crews arrive. What lessons are emerging from those early deployments?

The biggest lesson has been consistency.

One of the challenges with manual documentation is that it depends on people having the time to do it. On a busy project, it's easy for documentation to become a lower priority because everyone's focused on solving immediate problems.

Autonomous capture changes that. Robots can document the site on a predictable schedule, often overnight or before crews arrive, creating a consistent visual record every day without interrupting construction activities.

That consistency gives teams a much clearer understanding of how a project is progressing over time. It also makes documentation much more resilient. If a superintendent is tied up or a project engineer is out that day, the visual record doesn't stop. Everyone still has access to current information about what's happening on site.

Capturing a visual record has always been a natural byproduct of walking a site with a 360° camera. It's also a very natural task to hand to a robot. That allows experienced people to spend more of their time interpreting information and making decisions instead of simply collecting data.

• Looking ahead, what indicators should the industry watch to better understand whether construction robotics is moving beyond pilot programs and toward broader adoption?

I don't think the biggest indicator will be the number of robots on jobsites. It will be whether contractors keep using them after the pilot is over.

Construction is a very practical industry. Builders don't adopt technology because it's exciting, they adopt it because it saves time, reduces risk or helps projects run more smoothly. If a technology creates more work than it eliminates, it won't last.

We'll also know the industry has reached the next stage when robotics becomes just another way of collecting project information. Builders shouldn't have to think about whether data came from a person, a robot, a drone or another autonomous system. They should simply have access to an accurate, up-to-date understanding of what's happening on their projects.

Ultimately, I think the story is bigger than robotics. The real-world economy doesn't run on documents alone. It runs on physical places, physical assets and physical work. If AI is going to create value there, it has to understand reality, not just language.

That's why we say agents need eyes. When AI can reliably see, understand and reason about what's happening in the physical world, robotics becomes much more than automation. It becomes a new way for people to interact with the built environment and make better decisions. I think that's the transition we'll be talking about over the next decade.

• Northrop Grumman's MRV launched with two satellite-servicing robotic arms
• Mission Extension Pods can stretch a satellite's working life by eight years
• The spacecraft can inspect, refuel, tow, and relocate active orbital satellites or even destroy enemy satellites

Northrop Grumman has launched its first Mission Robotic Vehicle (MRV), equipped with two robotic arms to inspect, repair, refuel, and relocate satellites already operating in orbit.

The spacecraft also carries three Mission Extension Pods (MEPs) that can extend a satellite's operational life by as much as eight years after installation.

Although introduced as a commercial servicing platform, the spacecraft's ability to operate alongside other satellites has already raised questions about possible military applications.

Satellite servicing technology also enables close-proximity operations

The MRV launched aboard a SpaceX Falcon 9 rocket from Cape Canaveral, carrying MEPs intended for commercial satellites operated by Australia's Optus and Luxembourg's SES.

Northrop Grumman says the spacecraft can inspect satellites, install life-extension modules, tow spacecraft between orbital positions, and assist with clearing debris already circling Earth.

Those missions depend on two robotic arms that allow the vehicle to maneuver extremely close to other satellites without relying on direct physical contact.

The robotic arm system was developed through the Robotic Servicing of Geosynchronous Satellites program led by DARPA together with the United States Naval Research Laboratory.

Because those same capabilities allow precise movement around spacecraft, a future version could disable hostile satellites if governments requested such capabilities.

"I will leave it up to the U.S. government to decide how that capability might fulfill mission objectives that they have in that regard," said Northrop Grumman CEO Kathy Warden.

"It's our objective to offer our clients options for technology deployment, and for them to determine policy for when and how they might deploy that technology."

Military officials say orbital threats are already evolving

United States Space Force officials have repeatedly argued that rival nations already possess satellites capable of approaching and interacting with other spacecraft operating in orbit today.

The service says those developments require greater attention because several countries continue expanding technologies capable of operating near valuable military and civilian satellites.

"We're basically responding to a warfighting domain where our adversaries have already put interceptors in space, and we want to make sure that we rebalance that in terms of deterrence," said Chief of Space Operations Gen. Chance Saltzman.

A Space Force spokesperson later cited China's Shi Jian 21 satellite, equipped with a robotic grappling arm, while explaining examples of orbital interceptor capabilities already available.

Officials have also warned that close-proximity satellites could collect intelligence, interfere electronically, damage spacecraft components, or deliberately move another satellite from its assigned orbit.

Those evolving capabilities have prompted the Space Force to expand its focus on orbital warfare alongside traditional space operations.

The service's Space Delta 9 unit develops tactics for both defensive and offensive missions in orbit as those threats continue evolving.

The military is also planning larger space-based defense networks through the proposed Golden Dome missile defense architecture.

The Congressional Budget Office estimates the concept could require about 7,800 space-based interceptors if the program moves forward.

For now, the MRV remains a commercial tool, but how governments will apply comparable space security tech with AI tools and autonomous systems remains to be seen.

• Vodafone gives mobile masts autonomous antenna adjustment using artificial intelligence
• AI now rotates mobile antennas without waiting for engineering teams
• Robotic mast completes network optimisation in under thirty minutes

Vodafone has begun trialing a mobile mast fitted with an AI system and a robotic arm capable of adjusting radio antennas automatically.

The technology combines AI analysis with automated hardware movement to reduce manual intervention across 4G and 5G networks.

An algorithm developed internally by Vodafone first evaluates network demand and local conditions before directing the robotic arm's movements.

Robotic arm handles physical adjustments

The robotic component itself comes from Humax Networks, a South Korean equipment supplier working alongside Vodafone on this trial.

Once the AI algorithm identifies where coverage or capacity needs shift, it instructs the arm to rotate or tilt the antenna accordingly.

This process reportedly takes between 20 and 30 minutes to complete while optimising for energy efficiency across the network.

Manual antenna reconfigurations, by comparison, often take days or weeks once planning approvals, site access, and road closures are factored in.

Physical engineering works are typically required as well, adding further delay to any conventional antenna adjustment process.

The system adapts coverage based on environmental patterns, pushing more signal outdoors during clear weather and boosting indoor coverage when conditions turn poor.

Self-organising algorithms also respond to shifts in local traffic congestion, allowing antennas across several sites to reposition ahead of anticipated demand peaks.

Vodafone is currently testing this setup at a site overlooking a shopping centre in Tirana, Albania.

There, coverage automatically shifts between retail and residential zones as usage patterns change throughout the day, tracking real consumer movement.

Francisco Pignatelli, director of mobile access engineering at Vodafone, described the trial as an early example of physical AI within telecoms infrastructure.

“By analysing network demand and making autonomous decisions, the AI algorithm can adjust radio antennas to optimise coverage and capacity without human intervention,” Pignatelli said.

Internal antenna components signal the next shift

Despite the current trial's reliance on physically moving robotic hardware, Vodafone already anticipates a different approach becoming standard soon.

The company expects autonomous antenna technologies capable of adjusting internal components, rather than repositioning entire units, to reach the market within the coming months.

This shift toward internal adjustment is expected to reduce energy consumption further compared with moving an entire antenna assembly outdoors.

Vodafone stated that combining antenna digitalisation with AI-powered automation is a major step toward fully self-adapting wireless networks.

Whether internal antenna adjustment can match the physical precision achieved through robotic arm movement remains an open technical question for now.

Vodafone has not yet detailed how internal components alone will replicate the directional control that robotic rotation currently provides across varying environmental and traffic conditions.

Via Mobile World Live

• White Eagle knocked Matador's head off with a precise high kick during combat
• Matador continued fighting after losing its head and head-mounted sensors during battle
• EngineAI's humanoid robot can perform uppercuts, spinning kicks, and rapid recovery after falls

A white T800 humanoid robot named White Eagle stunned spectators in China after landing a precision kick that dislodged its rival's head.

The bout was part of the opening contest of the Ultimate Robot Knock-out Legend (URKL) tournament, the world's first full-scale humanoid robot fighting event.

Held at the Shenzhen Nanshan Cultural and Sports Center, the event drew 32 teams from around the world.

A robot that kept fighting

The most striking moment came when White Eagle's opponent, a black-clad robot called Matador, lost its head to a precise high kick.

Despite losing its head-mounted sensors, the machine kept throwing punches and absorbing strikes, finishing the match through its torso-based core systems alone.

Organizers said the sequence shows the robot's durability and its resistance to sustained physical impact, according to the Guangzhou Daily.

According to EngineAI, the T800 stands 1.73 meters tall and can perform uppercuts, spinning kicks and rapid recoveries after falls.

The company said each unit costs roughly $50,000 and features strong posture control, dynamic perception and shock-absorbing capabilities for combat.

All 32 competing teams used the EngineAI T800 as a standardized platform, ensuring every fighter entered the ring with identical hardware.

Selection for the inaugural tournament reportedly involved rigorous testing of each robot's stability and combat responsiveness before the event officially began.

The Shenzhen Nanshan Cultural and Sports Center provided the backdrop for a tournament organizers hope will grow into a recognized global sporting franchise.

Tournament tests more than robot strength

The URKL format assesses robots through effective strikes, body stability, defensive and evasive abilities, and overall durability in combat.

That structure means victory depends on more than simply delivering the hardest punch or kick.

EngineAI founder and CEO Zhao Tongyang said the tournament aims to develop a globally influential commercial humanoid robotics fighting franchise.

He also said competition could support research and industrial development by exposing machines to demanding physical conditions.

"We want to use competition to drive research and industry development," Zhao said. "Let the event feed back into technology, and let technology drive the industry."

Real combat could help assess mechanical balance, millisecond-level decision-making and coordination between multiple sensor systems.

Feedback from such conditions could accelerate the movement of intelligent robots from laboratory development toward commercial use.

Actor Donnie Yen attended the event and compared the experience favourably with staged combat scenes from films.

"Honestly, I used to see robot fighting only in science fiction movies. But today, for the first time, I got to watch real robots fighting up close,” Yen said during the opening ceremony.

The event's unusual head-loss incident reveals the durability of the T800, while also showing the limits and risks of using complex humanoid machines in physical combat.

Future contests will likely determine whether such events can produce useful engineering data beyond entertainment.

Via Global Times

Robot lawn mowers are growing in popularity, and with good reason. They're getting more advanced in features, easier to set up and use, and can be a huge time- and effort-saver... provided you pick the right model. There are plenty of options to choose from, spanning a range of prices and capabilities, and if you're new to the lawnbot world then it can be tricky to figure out which one you need.

This article is here to guide you through that process. I've pulled together five big questions to help you narrow down your options. And for added expertise, I've enlisted the help of KK Yin, Marketing Director at leading lawnbot brand Mammotion, whose insights you'll find throughout the piece.

#1. Do I need all-wheel drive?

"Everyone has a different environment for their lawn. So the first thing a customer should do is choose which kind of power system they need — for example, all-wheel drive or rear-wheel drive, or some the brands have front-wheel drive," says KK.

Rear-wheel drive will be perfectly fine if you have a relatively even and flat lawn, and will have no issue scaling relatively gentle inclines. However, if you have a very steep or uneven lawn, all-wheel drive (AWD) will help ensure the lawnbot maintains its grip without churning up your grass. Front-wheel drive is really only for flat lawns, and is a less common power system.

AWD tends to be reserved for larger, more premium robot mowers. However, at the start of this year Segway Navimow launched compact AWD model, with the suggestion that it would treat your turf more gently.

#2. Does my lawn have shaded areas?

Next, KK suggests figuring out which navigation method is best suited to the environment. "A customer should look at if they have shaded areas, the house is tall or they have trees in the garden. If they do, I suggest they don't choose an RTK positioning system. They should maybe choose a LiDAR model," says KK.

There are two major navigation systems for robot mowers. RTK, which uses satellites, can work very well if the lawnbot has a wide, uninterrupted view of the sky. However, if there are obstructions — such as those listed by KK — in the way, these can block the signal and lead to the lawnbot getting lost.

LiDAR is the system used by robot vacuums, and is relatively new in the robot mower market. LiDAR builds a picture of its surroundings by bouncing light off objects, so it's not well suited to wide-open lawns with no objects in them.

Broadly speaking, RTK will be a good choice in big, open yards, whereas LiDAR might be a better fit if you have a smaller, more cluttered yard. LiDAR models are often pricier, so if you can get away with an RTK bot, you might prefer to go with that option.

#3. How big is my lawn?

Many lawnbot models are available in a few different iterations aimed at different-sized lawns (typically indicated by a number after the product name). The difference is usually the battery capacity — a bot designed for a larger area will have a bigger battery, enabling it to cover more ground before having to return to its dock to recharge, thus making it more efficient.

"I think most users figure out that in one week they will mow maybe two or three times. So they need to work out how long it will take a product to mow their lawn once," suggests KK. With that information, you can work out if your mower will be efficient enough to keep up with your grass-trimming demands.

#4. Do I have more than one mowing zone?

"The next point, I think, is about multi-zone management," says KK. "Not all people have one big lawn — it's divided by the house, or the garden is split into different areas. That means they need to manage more than one mowing zone."

Pretty much every lawnbot will be able to mow multiple separate lawns, with the ability to create different maps in your app. However, some have more advanced multi-zone capabilities than others. One bot might be intelligent enough to handle a schedule where the lawnbot spends the morning in Zone A and moves to Zone B in the evening, and it might even be able to work out the optimal mowing patterns and cadence.

#5. How much setup do I want to do?

If you end up with a satellite-based lawnbot (see point #2), you might need to decide between a bot that comes with a separate antenna and one that does not. This RTK station is required to make the satellite information more accurate — so satellite positioning on its own might tell you where a bot is to within a meter or two, but add in a separate RTK receiver and it's more like centimeters.

Traditionally, you'd have this receiver in your own back yard. This adds an extra level of complexity to setup — the reciever needs to sit high up, with direct line of sight to satellites in the sky, and it also needs a power source. However, increasingly, we're seeing lawnbots that use a big, centralized RTK receiver that covers a large area. Mammotion calls this NetRTK. It's definitely a more low-effort option.

There are further features to look out for if you want the most straightforward setup. Some modern robot mowers have cameras and AI features that enable them to instantly create a map of their location, so you can just pop it on the floor and it'll be able to start mowing pretty much instantly. KK suggests these kinds of features are more of a nice-to-have, and advises focusing on the first four points first.

Reviews round-up

We've tested plenty of excellent lawnbots at TechRadar. Here's a taster of some of our favorites — tap the 'View details' button for more information, plus a link to our full reviews.

• A new 'brain-to-robot' interface has been unveiled in China
• It uses non-invasive EEG technology
• The system will supply much-needed real-world AI training data

We're seeing brain-computer interfaces (BCI) continue to get more advanced and more practical, and Chinese company BrainCo has now announced another first: a BCI that enables robots to be controlled by thoughts.

As reported by the South China Morning Post, BrainCo says the system works through EEG (electroencephalogram) headsets worn by human operators, which translate neural activity into instructions for robots to follow.

Like other similar 'mind reading' devices, it's based on vast amounts of training data that associate specific neuron activity with specific commands as they're imagined in the brain. AI algorithms then manage the job of finding the right action for the right thought.

It's called the Brain-Controlled Robot AI Platform, and it was unveiled at the World Artificial Intelligence Conference (WAIC) in Shanghai —the biggest and most important AI tech event on the calendar in China.

Think it into reality

BrainCo says that as well as being used to control robots, the system can also act as an interface to connect to robotic arms and other devices. A robotic arm could be instructed to pick up a cup through thoughts alone, for example.

As the Brain-Controlled Robot AI Platform becomes more widely used, it will also be collecting data — data that can then be applied to refine the underlying algorithms further. High-quality training data is currently one of the major bottlenecks in the industry.

"The embodied AI industry has made remarkable progress on what robots can do on their own," said Nyx He, a partner and senior vice-president at BrainCo. "We believe the next decisive frontier is about how robots understand the humans they work with."

With AI models continuing to get smarter and more sophisticated, this is going to feed into the 'physical AI' of robotics — giving these machines greater flexibility, more dexterity, and as we've seen here, thought-controlled inputs.

• Ukraine's Brave1 opened the first state grant program treating combat humanoid robots as their own procurement category
• The move, which aims to capitalize on gains by Ukraine's already-innovative robotics industry, is also cognizant of the current limitations of the tech
• The goals set by Brave1 are modest but offer backing of up to ₴100M+ for breakthrough projects

Ukraine has become the first country to fund combat humanoid robots, but the move might prioritize future innovations over deployable solutions in the near future.

The devil is in the details: its own battlefield data shows wheeled and tracked bots running 66,000+ missions in 2026 so far, while the only combat-tested humanoid lasted three hours on a charge.

This points to the much humbler beginnings that Ukraine's own moves in the space indicate, even as it continues to scale up its bot deployments in the field.

What is Ukraine actually buying in 2026?

Ukraine's move comes in the form of the Brave1 Advantage event in Kyiv on 2 July 2026, where company CEO Andriy Hrytsenyuk announced a grant competition for domestically built bipedal humanoid robots designed exclusively for military tasks.

The move, spearheaded by Brave1, is part of Ukraine's military push for innovation on a fast-changing battlefield; winning designs can receive over ₴100 million (roughly $2.4 million) in grants.

Brave1 was founded in April 2023 by six ministries and is now being transferred from the Ministry of Digital Transformation to the Ministry of Defense; it has become Kyiv's main mechanism for converting frontline needs into funded engineering projects.

CEO Andriy Hrytsenyuk painted this as a must-have industry for Ukraine, stating, "We see how quickly the humanoid robotics industry is developing worldwide, in China and the United States. We see that such robots have value for strengthening our military capabilities. That is why we are moving in this direction."

The move seems to be a long-term play on robotics, with Defense Minister Mykhailo Fedorov also weighing in earlier using three words: war must become a format where "technology fights technology."

The reason one should expect Wall-E rather than Terminator, however, is that Ukraine already runs the largest ground-robot ecosystem in any active war, and none of it walks on two legs.

Ukraine has also made significant inroads, such as holding off Russian attackers at key positions and capturing an entire fortified Russian position with the help of drones and ground robots, marking the first full seizure of territory in the world by bots.

For now, the only humanoid that has actually been to war in Ukraine is the Phantom MK-1, built in San Fransico by a startup, Foundation Future Industries.

The MK-1 carries only about 20 kg, has no waterproofing, and provides roughly two to three hours of battery life, compared with infantry missions that run eight to twenty-four hours, rendering it somewhat ill-suited to most of the combat operations that Ukraine currently conducts along the frontline.

Ukraine's investment in humanoids might not, therefore, result in short-term fruitful gains but rather indicate it is bucking the trend, so to speak, even as Goldman Sachs is projecting nearly 50,000 to 100,000 non-combat humanoid shipments globally this year, even as military planners envision putting one in a trench in the near future.

• Neo Beta robot hands have human-level dexterity
• They use tendon-like controls instead of in-hand motors
• They're waterproof and can perform many skills normally reserved for human hands

Neo's new robot hands are so good and lifelike, you might assume that they're gloved human hands, and then you might wonder why someone is swinging a hammer at them.

In a new demo video recently released on YouTube by Neo Beta robot parent company 1X, you can watch a pair of Neo Hands screw in a lightbulb and pull the chain switch (before someone inexplicably shatters the bulb with a hammer), pluck grapes off a bunch and drop each one in a container, carefully pickup a screw, unzip a jacket (a bit creepy), and even open a small bag of Funyuns onion ring snacks.

In that last bit, someone swings a hammer at the hands while they work, which they pay no mind to, before the bag is unsealed.

It even expertly builds a small Lego stack. Okay, okay, they're the larger Duplo blocks, but it still does as good a job as your average kindergartner.

The hands move slowly but also with a grace and ease you might mistake for humanness. This appears to be down to the underlying technology.

As 1X described it, the rubber-covered, waterproof hands use a "closed-loop tendon-driven system". This means that 1X moves the motors or servos out of the hand and back along the arm, which keeps the hand smaller and more supple. Those motors are then connected to an intricate system of tendon-like connectors that are pulled and released to enable the robot hands' movement and manipulation. That style of control more closely resembles our own hands, which, while including muscle, are also filled with tendons that are pulled inside the forearm.

1X says the fingers, palm, and thumb have 25 degrees of freedom, but, as evidenced in the video, they can also over-extend backward in a rather unnatural or certainly more-than-human way.

The robot hands feature some impressive strength, too, lifting a 20 lb. dumbbell and then, with just one finger, curling a smaller pulley weight.

Naturally, the fingers include sensors so that the hands and robot know when it's gripping something and how much force is or should be applied. That's how the Neo robot hand avoids breaking that light bulb (unlike that hammer).

Next-level grace

I know, just last week, we heard about a pair of robots performing gall-bladder surgery on a pig. You'd think that those humanoid hands must've been far more graceful. However, those robots each simply gripped a pair of laparoscopic controls so that the finer control, where the cutting and suturing was done, happened at the end of those devices. No robot hand directly manipulated the scalpels.

While Neo's robot hands will be useful for all kinds of home helper chores when the $20,000 Neo finally arrives in consumer homes (early adopters may be receiving them now), the hands are also useful for helping itself: In the video, a Neo Beta robot uses the hands to pick up its MagSafe-style charging puck, which it carefully attaches to its robot hip.

1X writes that these hands will deliver new real-world training data to its robotic development, and one would assume this will ultimately make the Neo robots even better home helpers, companions, and enablers (all those Funyuns).

• UN Secretary General calls for a global ban on autonomous "killer robots”
• Guterres argues that delegating life-or-death decisions to machines is “morally repugnant”
• Governments should take a stance now – not wait for something catastrophic to happen

UN Secretary General António Guterres has called for lethal autonomous weapons, which he describes as ‘killer robots’ to be prohibited under international law following recent discussions at the first Global Dialogue on Artificial Intelligence Governance in Geneva.

Guterres’ demand to ban these weapons focuses on those capable of identifying, selecting and attacking targets without human oversight, which leaves artificial intelligence and other computer systems in charge of a life-or-death decision.

He ultimately argued that certain decisions must remain exclusively human, and the decision to take a life is well into the boundary of requiring human oversight. Transferring the decision-making to killer robots would be “morally repugnant” and “politically unacceptable,” he argued.

AI requires global regulation as military AI poses major threats

Key to the Secretary General’s argument is that he urges governments to take action and ban such robots now, rather than waiting for an autonomous weapon to cause a major incident before rethinking their strategies.

“Let us not wait for atrocity to act,” Guterres said. “Some decisions must remain forever human – none more than taking a human life.”

The issue is becoming more urgent now that AI models and advanced chips are already being used within military intelligence, targeting and other battlefield systems.

More broadly, Guterres’ thoughts align with those of Anthropic, which recently had a dispute with the Pentagon after seeking guarantees that its models would not be used for autonomous weapons or surveillance.

While the Pentagon had rejected those limitations, arguing that it should be able to use Anthropic’s models for any lawful purpose, the case highlights how private companies are becoming increasingly intertwined with digital warfare.

Reporting by the Wall Street Journal cited a similar view by Pope Leo XIV, who warns that AI-controlled weapons could promote an “anti-human” view of warfare. He warned that the autonomy could reduce some dangers and distance political leaders from the human consequences of conflict.

There’s a need to balance the pros and cons of AI

However, artificial intelligence does promise several benefits to modern warfare, particularly in its ability to process huge amounts of information extremely quickly. With modern compute, militaries can respond to threats at lightning speed, improve their accuracy and precision, reduce soldier risk and potentially reduce civilian casualties, too.

Critics also question whether human oversight of AI systems is at all meaningful if the person in charge only has seconds to act on AI-generated information in the first place.

It’s also yet to be determined which party or group of parties should be held accountable for any incidents or mishaps – human operators, commanders, hardware manufacturers and software developers are just some of the parties up for judgment.

“We may be the last generation able to set the terms on which humanity and machines coexist,” Guterres warned separately in an X post, warning that AI must be governed, trusted and fair.

“It sounds like science fiction, but it's a real possibility, and it could change the world in ways that we don't understand yet, and it could change the power dynamics of our planet in ways that require our attention,” Independent International Scientific Panel on AI Co-Chair Yoshua Bengio added.

Chinese robotics firm Agibot has launched a range of humanoid robots designed to look, move, and sound like human beings in the UK business-to-business (B2B) market.

The Shanghai-based company, founded in 2023 by two former Huawei engineers, hosted a launch event in London on June 30, 2026, with several of its A3 full-sized humanoid and X2 half-size humanoid robots in attendance, as well as current and prospective business partners.

The robots — all controlled by humans with smartphones and DualSense controllers — were deployed at several demo stations in the event space.

One X2 humanoid robot was set up to dance to peppy music, cutting some serious shapes for a constant crowd, while another in a Harry Kane shirt kicked a football into a net. A third wore a suit of (admittedly very cool) Gundam-style armor, demonstrating the robots’ potential for customization.

As for the full-sized A3 humanoid — less expressive than its smaller sibling, with sharper angles and a rather blank circle in place of the X2’s animated ‘eyes’ — a unit was made available to greet guests and for photo opportunities.

In a presentation at the UK launch event, Agibot co-founder and COO Daniel Jiang said: “The robot business is the AI business. Robots should understand you, know what you say, talk to you, and work for you.”

Indeed, at the launch event, the word ‘robot’ was as common as the phrase ‘embodied AI’, as Agibot representatives attempted to position the firm’s product range as a signifier of a new technological era: the ‘fourth industrial revolution.’

In a later talk, co-founder and R&D president Yan Xiong said: “Of the products I’ve worked on, embodied AI is the most exciting and inspiring. Through three intelligences in one body, robots can become human-like.”

The three intelligences in question control interaction, locomotion, and manipulation, powered by on-device AI models that communicate with cloud servers. It seems that Agibot sees itself as an AI company first and foremost, and is allocating its resources in kind: “Our main focus is the research and development of the three intelligences, taking up 75% of our R&D staff, and 80 per cent of our R&D budget.”

That’s not to say that the firm's humanoid robots aren’t profoundly impressive. The units are made mainly of magnesium alloy, with titanium reinforcements at pressure points to ensure rigidity. The A3 weighs 55kg, which is pretty reasonable when you consider its 5-foot-8-inch (173cm) height, and lasts up to 10 hours on a single charge of its 1152W battery.

In comparison, the 4-foot-3-inch (131cm) X2 weighs 39kg and lasts for two hours. The units communicate with the internet via dual-module, dual-SIM 5G.

Also present were units from Agibot's D1 quadruped range, inexplicably programmed to do backflips on command, as well as G2-series industrial robots and C5 cleaning robots. The top-tier D1 model, the D1 MaxPro, has a carrying capacity of 50kg, while only weighing 68kg itself.

Agibot is young but quickly increasing its production capacity. It recently rolled out its 15,000th unit, having passed 10,000 units just three months earlier, and has already deployed its G2 industrial robots in China’s Longcheer electronics factory. Indeed, much of the event was dedicated to promoting possible use cases for each model in industries like construction, live entertainment, and the corporate world.

Having shaken hands with the X2 and watched the D1 wave its ‘paw’ and, again, perform several perfect backflips, I have to admit that I’m still stunned by Agibot's range of robots. The firm’s launch has done nothing to allay my many concerns about AI, but it’s still almost unbelievable to see actual droids walking around.

It’s clear that Agibot wants its robots to be relatable — the dance moves, handshakes, and free kicks are obviously an effort to instill some kind of affection in a crowd of possible customers. I’d be lying if I said I didn’t find the X2 wildly entertaining, even though I know it's just metal and plastic — but that’s the power of pattern recognition.

If Agibot can make strides towards its professed goal of autonomous operation, then we really aren’t too far away from having a real-life C3PO on our hands.

It’s early days, but the firm’s leadership sounds committed. In his presentation, Yan Xiong said: “Scaling to mass production is the first hurdle to mature the embodied AI industry. After we achieve mass production, the next step is to bring our portfolio to commercial availability."

“AI should not be confined behind screens,” he later added, “it must be brought into the physical world.”

• Japan plans to deploy 10 million additional robots before 2040 arrives
• Nursing homes and food factories sit at the centre of expansion
• Noetra will supply the AI foundation powering future Japanese robots nationwide

Japan has unveiled a revised national robotics strategy which aims to introduce around 10 million robots in the country by 2040.

Economy, Trade and Industry Minister Ryosei Akazawa announced the plan, which now spans 18 fields after adding food manufacturing and medical care to earlier priorities.

The government intends to move quickly to establish a core AI robotics hub, supporting deployment, research, and workforce training activities across the country.

Robots move beyond the factory floor

Officials described the hub as central to helping companies adopt robots at scale over the coming years, particularly in sectors already struggling with staff shortages.

The strategy centres on Noetra, a domestically produced multimodal foundation model developed alongside a National Institute of Advanced Industrial Science and Technology project focused on physical AI.

Noetra is majority-owned by SoftBank, NEC, Sony Group and Honda, while Fujitsu and Rakuten are reportedly still weighing whether to join the consortium.

Akazawa said accumulated data from elderly care, disaster response, manufacturing sites and the Fukushima Daiichi decommissioning effort underpins the government's confidence in the approach.

"The utilization of accumulated data" would become Japan's "winning strategy," he said, framing global competition as a contest over accessible datasets rather than raw computing power alone.

The government plans to build data infrastructure for physical AI and robots that reflects the country's own industrial strengths.

It will lean on decades of experience operating machinery in hazardous or labour-scarce environments nationwide.

International partnerships and regional ambitions

Officials have confirmed a collaborative arrangement with research institutions in the US, Canada, France and the UK to support development of the base model.

The resulting technology will reportedly be made available widely to Japanese AI developers, businesses and eventual users across multiple industries and regions.

According to officials briefed on the plan, some companies are expected to use the platform as a foundation for expanding into foreign markets in later years.

The minister also linked the strategy to broader efforts encouraging AI-driven transformation originating from regional areas outside Japan's major metropolitan centres, rather than concentrating growth in Tokyo alone.

Japan's ageing population and restrictive migration policies continue creating labour shortages across industries struggling to recruit sufficient numbers of workers.

Policymakers increasingly view automation as a practical response because many vacancies remain difficult to fill through conventional hiring efforts alone.

Supporters frequently argue that robots are filling roles unavailable to human workers rather than directly replacing existing employees across industries.

The revised strategy therefore includes medical care responsibilities alongside duties within food production and beverage manufacturing environments across the country.

South Korea announced a comparable robotics ambition this week, adding a competitive dimension to the wider regional picture as both countries pursue sovereign AI capabilities.

Whether those ambitions become reality may depend less upon announcements than sustained investment, technical progress, and broader public acceptance domestically.

Via The Register

• A Chinese robotics firm has said that some workers ‘want to be replaced’ by its humanoid robots
• Agibot has launched its range of humanoid and quadruped robots in the UK B2B sector
• The company's Europe and US lead also said that robots could become nurses and teachers

An executive for Chinese robotics firm Agibot has said the company believes its humanoid robots could replace certain human workers — and that “in some jobs, they want to be replaced.”

Agibot, founded in 2023 by two ex-Huawei engineers, makes humanoid, quadruped, and cleaning robots for the business-to-business (B2B) sector, advertising use cases like manufacturing, cleaning, entertainment, and construction.

The company recently rolled its 15,000th unit off the production line and announced its expansion into the UK market at an event in London on June 30.

In a press conference at the event, attended by TechRadar, the president of AgiBot Europe and America, William Shi, told reporters that “we’ve got to factor people doing dangerous, boring, and repeatable jobs — these kinds of jobs can be very easily replaced.”

Shi added: “For some job descriptions, they want to be replaced, because it’s very boring, very dangerous, and very high-risk — nobody wants to do this [kind of job]”.

‘Replaced by robotics’

Agibot's product range includes the full-sized A3 humanoid robot, the half-sized X2 humanoid robot, and the D1 range of quadruped robots (with form factors resembling dogs, though the company never describes them as such).

Though the company has just landed in the UK, its robots have been successfully deployed in manufacturing facilities in China. A recent YouTube livestream shows the G2 industrial robot ‘at work’ at the Longcheer electronics factory, where the humanoid robots are deployed alongside human employees.

Speaking about Agibot's deployment with Longcheer, Shi said: “They have a lot of workers standing for eight hours per day. They take a smartphone, turn it around, and put it in a box. And then they take the box and move to the next production phase.”

“These kinds of steps can be easily — and are expected to be — replaced by robotics,” he continued, “because that doesn’t create value or create happiness for people. They don’t learn when they do this job. They don’t invent things.”

At present, Agibot's products are not autonomous, but each is equipped with a three-part AI model, with each part controlling interaction (with people), locomotion (moving around), and manipulation (of the local environment — i.e., picking stuff up), respectively.

The company is actively aiming for autonomy, powered by the collection of usage data and further AI development, but Shi emphasizes that Agibot humanoids will remain “under the control and expectation of the human.”

Robot child-rearing

As well as manufacturing, Shi highlighted baristas, live entertainment workers, and even teachers and nurses as professions that could be occupied by robots: “There’s a big lack of nurses in every country, from China, to America, to Europe, and also a big lack of teachers.”

“Most of children’s questions can be responded to [by robots] — phonics, science, and mathematics questions, or even some common conversations. You ask for the weather, the humidity; you can ask these questions to all the robots, because they’re based on large language models.”

Still, there’s some distance between using ChatGPT to conduct research and allowing a humanoid robot to teach children — and according to 2025 research by KPMG and the University of Melbourne, while almost three quarters of UK adults use AI at work, less than half say they trust it.

With the B2C market a distant target for Agibot and an array of business partners — including Nvidia, which provides chipsets for the robots — backing the company's UK launch, it's clearly expecting to make big progress in B2B industries first and foremost. But Agibot isn't ruling out a future in which robots are part of our everyday lives.

“What we want in the future is that the robot can take some responsibility in daily life,” said Shi, “but they will never make the decisions instead of a human.”

• Weave Robotics Isaac 1 is now on pre-order
• It promises to autonomously handle basic household chores
• The roughly $8,000, legless robot arrives in the US first

With its Baymax-esque face, cool hues, and mostly soft body rolling on a wheeled base, Weave Isaac 1 cuts a distinct figure from most humanoid robots. It's just cute/homely enough to attract the robot-curious while avoiding the dangerous uncanny valley of, say, UBTech UWorld U1.

That's probably a good thing. Weave, which has been building robots like this for a few years, is finally ready to fully commercialize its offerings with the $7,999 (around £6,000 / AU$11,600) Isaac 1. Far from the overreach of many AI-backed bots that promise not only to handle home chores but also to be your empathetic companions, Isaac 1 has smaller goals.

Its features can be boiled down to household drudgery you probably hate to do, including:

• Laundry
• Finding and picking up "dirty" clothes (it's not clear it knows how they are dirty)
• Hamper duty
• Folding and putting away the clean clothes (missing here is setting up and running the washing machine or dryer)
• It can make your bed
• Fluff the pillows
• Clean up the clutter (shoes, toys, etc)

It's a good list, but despite Isaac 1's autonomous capabilities, also limited. So much so that Weave Robotics promises teleoperation to help it complete some tasks. For anything you want it to do, you launch it in a connected smartphone app. After that, it rises up from a 3ft tall crouch at its charging base to its full 5 ft 9 in height and rolls out to get the job done.

In the launch video, Isaac appears to move slowly and methodically, but it might be a bit steadier than, say, the likes of Tesla Optimus, Neo Beta, Unitree G1, and Figure 03, all of which are built to replicate human bipedal walking. Those robots get better and better at walking, running, and dancing, but, as we've seen, they can be more unpredictable.

In addition, all those extra actuators and sensors required for walking tend to make these bots more expensive (they can run from $12k to $ 20k or more). With its wheeled base and sensors to autonomously navigate your home, Weave's Isaac 1 appears steadier and is far cheaper than many of these potential home helper or companion robots.

While Isaac 1 is ready to fold your laundry, it will do so without the use of fingers. Instead, Isaac 1 will manage laundry and other chores with a couple of orange claws (and reportedly run for eight hours on a charge). The rest of the body is covered in soft, removable body plates in muted colors like Sage, Gray, Slate Blue, Terracotta, and Vesper.

Like other robotics companies, Weave Robotics is offering a subscription plan: $449 a month (which equates to about £340 / AU$650 a month), which you would pay in perpetuity, but that might also ensure perpetual updates. Getting one, though, starts with a $250 deposit, and deliveries are expected to begin in September, but only in California. Other areas in the US will follow, according to the company, in 2027.

Isaac 1 won't ever be mistaken for a human, but perhaps that's the point. In the rush to recreate and automate humanity, most robotics companies seem to have forgotten that humans respond best to other real, flesh-and-blood humans and, at a fundamental level, reject artifice and hardware that resembles but can never be us.

• UBTech just introduced its most-human-yet robot
• The silicone skin is realistic and creepy
• It launches in China this year, but there are also plans to customize it to look however you wish

Ubtech, last seen deploying humanoid robots to patrol the China-Vietnam border, is back with its creepiest creation yet, the UWorld U1, its first mass-produced, full-sized "ultra-bionic humanoid robot."

I know that's a mouthful, but the UWorld U1 Series does mark a departure for UBTech, which has traditionally produced faceless automatons aimed primarily at business, enterprise, and the border. However, the company has long had consumer ambitions, and the UWorld U1 Series may be the bot to realize them.

This new humanoid robot features silicon skin, a lifelike face with eyes that follow you, lashes that blink coquettishly (ick), and 88 degrees of freedom across its full-sized robotic body. It even has what UBTech calls a "dual-pivot biomimetic cervical spine," which apparently gives it more human-like movements.

UBTech unveiled the robots earlier this week at a splashy launch event in Shenzhen, China. There are three models in the Series: the U1 Lite semi-torso edition, the high-performance full-body U1 Pro, and the high-dynamic full-body U1 Ultra.

During the launch on June 30, 2026, several UWorld U1 Ultra humanoids walked the stage, interspersed with actual humans. The idea, I think, was to confuse the viewer so they might not know which was which. In truth, the robots all looked a little plastic, or like characters straight out of your favorite anime, and they walked somewhat awkwardly.

To further prove their believability, UBTech had one robot dance with a tuxedoed human. At times, it looked as if he might be supporting a potentially teetering UWorld U1 robot.

What are these robots for?

In addition to questionable dance moves, UWorld U1 is equipped with an "emotion-aware LLM" that the company claims will help it recognize and respond to "fine-grained emotional states." It's designed as a proactive companion, responding to human interaction in a fraction of a second.

UBTech also, somewhat comically, promises a focus on privacy, in a country where the Chinese government can ask to see all your data at any time. Still, UBTech's privacy architecture is smartly focusing on "minimal cloud dependency," which means most of your data may not be with UBTech anyway.

According to a release, the UWorld U1 is designed to meet a real and growing need in China, where the company claims that 90 million adults are living alone and there are 118 million empty-nest senior citizens.

The company is so concerned about this growing issue that it plans to donate 100 of these robots in 2026. But here's where things get really weird. I'll let Ubtech speak for itself:

"These units will incorporate 3D facial reconstruction and voiceprint-based identity replication technologies to recreate designated individuals, while integrating emotion-driven interaction models and dedicated long-term memory systems."

As I read that, they plan to make these robots look like the person of your choice. Perhaps it's a dead husband or maybe an adult child who moved away. The custom UWorld U1 Ultra will look, sound, and maybe even respond like your missing companion.

Yikes! How did we get here? Why is life imitating Black Mirror? In truth, these robots are unlikely to fool anyone into thinking they're really human. Sure, UBTech did its best to confuse us in the presentation, but in the real world, that cold silicon, odd gait, and sure-to-be-creepy interactions won't be fooling anyone and will be a poor substitute for your absent partner.

Also, if you really are interested, you'll have to move to China and plop down the equivalent of almost $18,000 in US dollars (around £13,570 / AU$26,120) when they ship sometime this year.

The rise of robotics has long threatened human employment, with machines in manufacturing settings especially seemingly performing the work that people used to do. Take this to its logical conclusion, and it's not outside the realm of possibility to have a massive unemployment problem on your hands. So how do we deal with it? Microsoft co-founder Bill Gates has a contentious idea.

Death, taxes and robots

There's an active (and ongoing) debate about how the rise of apparent job-stealing technology could be handled by policymakers.

This is a challenge to which Gates offered a solution nearly a decade ago, suggesting in an interview with QZ that a robot that performs a human's job should be taxed like its fleshy counterpart.

This, Gates argued at the time, could slow down the rapid speed of automation and be used to instigate exploration into other kinds of employment.

This is, of course, not a policy that everyone (especially much of the business world) agrees with. The International Federation of Robotics (IFR), for instance, decried the move and instead suggested further tax on profits to offset the impact of automation.

Tax-paying robots

Despite that, there are supporters of similar measures, including Anthropic's CEO Dario Amodei, who called for an AI tax proposal to build a multi-million-dollar worker fund. Andrew Yang, too, has called for a tax on automation and AI instead of a tax on humans.

Since 2017, the conversation has certainly moved on – with much of the conversation around automation based on software and digital tools like large language models (LLMs) rather than physical embodied AI robotics systems. That said, the robotics industry is growing fast. While AI is a key focus among policymakers, the conversation may well shift back to machines that we can see and feel.

• Alibaba's Tongyi Lab unveils the Qwen Robot Suite
• Its first embodied-AI models are split into navigation (RobotNav), a video "world model" (RobotWorld), and manipulation (RobotManip)
• The move comes after Nvidia recently unveiled and published its own Cosmos 3 offerings

As much of its AI competition continues to focus on LLMs and making them faster and more capable, Alibaba might be looking to lead on another frontier altogether, along with its LLM ambitions in tow: robots.

The company's Tongyi Lab has unveiled the Qwen Robot Suite, what it calls a family of models focused on "embodied AI," which centers on enabling machines to perceive space, reason, and act accordingly.

This comes at the heels of Nvidia's own Cosmos 3, a frontier model for physical AI, further bolstering CEO Jensen Huang's narrative that China's developer ecosystem remains relatively unaffected by chip restrictions, even as focus in the West continues to shift to power for many of the sprawling data centers being built in the US.

A competitor or a complement to Nvidia's playbook?

The Qwen-Robot Suite consists of three core models: Qwen-RobotManip, a generalizable vision-language-action model; Qwen-RobotNav, a scalable vision-language navigation model; and Qwen-RobotWorld, a video world model designed for embodied intelligence.

There is no denying, however, that robotics is being treated as perhaps the most crucial frontier for AI, even as LLMs continue to advance, with both Google and Nvidia among the companies pouring billions into research on their respective Gemini Robotics and open source Cosmos offerings.

Alibaba claims that the model, which leverages a more lightweight Qwen3.5-4B model rather than its Qwen 3.7 Max, which features over a trillion parameters, manages to top the RoboChallenge real-robot benchmark, scoring an impressive 59.83 and a 45% task success rate.

With other interested parties such as Tencent, Unitree, AgiBot, UBTech, Galbot, Spirit AI, and GigaAI, in addition to interest from EV firms including Xpeng and Xiaomi, all shaping the future of Chinese AI robotics, R&D in the industry is continuing at full swing, even as upcoming IPOs are expected to further propel the industry forward with easier access to capital.

The South China Morning Post, a wholly owned subsidiary of Alibaba, noted that "Alibaba’s entry comes as embodied intelligence is fast becoming the next frontier in global AI."

Nvidia's position on the matter is perhaps more nuanced with it attempting to behave as an 'enabler' versus a direct competitor as it pushes its open-source model to perhaps form the same building block that CUDA does for GPUs with Cosmos, GR00T, Isaac and similar offerings being the playbook this time around to ensure future robotics platforms are built, much like most AI tools, around Nvidia's hardware and software stack.

Alibaba's announcement might not be a sign of the Chinese giant out-engineering Nvidia, but in the backdrop of the Chinese government insisting informally at least, on a decoupling or at a minimum, no reliance on US-based hardware or software, it can be seen as an intent to build a similar ecosystem for Chinese robotics companies.

In the absence of Nvidia's presence in China, it might be hard to compare the two offerings, even as their scales differ considerably: Cosmos 3 is an open-world foundation model with multiple vendor-reported scores that do not cover RoboChallenge, whereas Alibaba's are self-reported from exactly one benchmark. Until both approaches can be compared directly, one can not assume superiority of one over the other.

What is, however, perhaps painfully obvious for Nvidia, and has been warned of time and again by its CEO, is that China, irked by US policies around AI, is no longer looking for chips, models, or even open source solutions to incorporate into its ecosystem, but wishes to build them from the ground up.

This could result in a lack of exposure to what was the second most lucrative market for the chip designer, a move that could cost it billions of dollars in revenue in the robotics segment alone, from what is still widely considered the "factory of the world" due to its huge manufacturing base.

• Report finds public support for workplace robots is growing – especially for physical or dangerous tasks
• Willingness to accept a robot inside the home grows with real-life exposure
• Familiarity and clear governance are essential to overcoming public fear

Technological capabilities may no longer been the limiting factor when it comes to how and where robots can be deployed, with new Hexagon research revealing public support isn’t always there.

The company found much of the public is becoming more accepting of robots in the workplace, but only where they’re used for practical, physical or dangerous jobs.

However roles which require empathy, judgement or human interaction are still where support remains low.

Robots are most accepted in practical labor use cases

For example, more than half (56%) of the 1,000+ UK adults surveyed said they’d accept robots in lifting and transporting heavy items. Carrying and delivering any items (38%) and monitoring hazards and dangerous environments (34%) also received reasonable support.

With airports, some supermarkets and other public places now employing robots, 31% would even support their use in cleaning shared spaces.

Though the research fails to detail perception by age bracket, the company surveyed an equal number of UK children to reveal that heavy lifting, carrying and delivering is even more accepted among under 18s.

However, while repetitive physical work is generally well-accepted, 82% of UK adults want humans to care for sick, elderly and young people.

Only 5% say they’d choose a robot caregiver, making this the lowest support for any of the tasks included in the report. Even children seem reluctant to have non-human personal interactions, with 79% preferring human caregivers and 8% willing to choose a robot instead.

But Hexagon Technology Ethicist Dr Blay Whitby argues a simple reframing could skew these figures: “Ask people if they want to be cared for by a robot, and most say no… Ask if technology should help them remain independent in their own home for longer, and most say yes.”

Associate Professor in Moral Psychology Dr Jim Everett sees robots more as “assistive devices” in care homes and classrooms, rather than human replacements.

Exposure can drastically shift public perception

For now, the public still sees robots as industrial automation roles. More than half agree their natural homes are factories (53%) and warehouses (53%) – fewer consider them at home in hospitals and clinics (34%) or classrooms (30%).

Fear of the unknown could be another blocker, with only 28% of UK adults believing that having a robot colleague would be exciting – nearly half (46%) say it would be frightening. Humanoid forms are clearly unsettling, with twice as many preferring machine-like robots (27%) compared with human-like robots (14%).

Sci-fi fears about robots taking over could also be influencing public perception. Nearly all UK adults (88%) want clear rules governing what robots can do.

“Industrial environments are where the tasks for robots are the most defined, the safety cases are mature, and governance is in public view,” Hexagon CTO Burkhard Boeckem concluded.

Global comparisons back the fear of the unknown theory – while 30% of UK adults have encountered robots in real life, 75% have in China. A country that’s nearly twice as likely (63%) to accept robots into the home compared with the UK (32%).