Microsoft's smart laptop learns from you to boost battery life

Using the Nest learning model

Desperately trying to find a power outlet to charge your laptop in the middle of a meeting or presentation may soon be a thing of the past. Instead of waiting for the next breakthrough in battery technology, researchers at Microsoft are working on a way to improve battery stamina using existing technologies.

This means that a future in which your dead laptop battery could quickly charge up in time for your next meeting, or having a laptop maintain enough battery life to last you through a transAtlantic flight, may be ready sooner than you think.

To create this perfect battery, Microsoft is using a variety of different types of battery, machine learning and smart software.

Current battery technologies

Battery advances have not kept pace with improvements in computing components, such as faster processors, higher resolution displays and faster wireless speeds.

Today, battery charging on a laptop is controlled by hardware rather than the operating system, which means that it is not as efficient. Hardware alone cannot determine, for example, how a user is using their laptop and in what context or setting to give the proper charge needed.

Microsoft researcher Ranveer Chandra points out that people use their devices for different tasks, each with different battery consumption needs. For example, viewing web pages and editing Word documents may have less battery impact than editing videos or performing complex data analysis.

Batteries in computing devices should address these different needs, Microsoft claims. Microsoft's goal is for the laptop to understand a user's immediate needs and provide a fast or more sustained charge.

Taking the software approach

By moving the battery charging controller from hardware to the operating system, Microsoft is taking a machine learning approach, not unlike what Google-owned Nest is doing for connected homes. By learning when users perform certain computing activities, the OS will be better able to determine what kind of charge is needed.

In addition to controlling the charging from a software level, Microsoft is also using different types of batteries to optimize power delivery in the research prototypes. Because no single battery can deliver the requirements of fast charging, high capacity, low cost, less volume, light in weight, less heating, better longevity and high peak discharge rates, different batteries are combined to address these different requirements.

"It combines several different kinds of batteries, all of which are optimized for different tasks, into the same computer," Microsoft said in a blog post. "Then, it works with the operating system to figure out whether the user is, say, looking at Word documents or editing video footage, and applies the most efficient battery for that task."

Touching the Surface

In addition to Chandra, Julia Meinershagen, Senior Engineer with the Surface Devices group, also participated in this research. If Microsoft is successful, we may soon see significantly better battery life on Microsoft's future generation of Surface devices.

But the innovation doesn't stop there. The same principles can also be applied to smartphones, cars and other gadgets, Chandra wrote. Currently, chipmakers like Qualcomm and Intel are relying on smaller architectures to deliver faster performance and longer battery life.

The best part about Microsoft's battery innovation is that it uses existing technology so we don't have to wait for the ultimate battery to be invented.