Microsoft underground

When we think about Microsoft, all sorts of things come to mind. Word processors, spreadsheets and servers, for example.

But wheelchairs? Seriously?

If you step into Microsoft’s underground, anything is possible.

The company recently gave a small group of journalists a tour of its labs and a chance to chat with researchers to offer a taste of what the future may hold.

Eye Gaze wheelchair
Eye Gaze wheelchair

The wheelchair project grew thanks to former NFL footballer Steve Gleason, who suffers from ALS, sometimes called Lou Gehrig’s disease. He asked for help in doing three things: moving on his own, playing with his son and talking to his wife naturally. For an ALS patient, who has little or no control of his body, that’s a tall order.

During a company-wide hackathon, a Microsoft team came up with a wheelchair that is controlled by simply looking in the direction you want to go on the screen of a Surface tablet. We tried it – and it’s a good thing that you can stop by simply closing your eyes. It takes practice to navigate, but it works astonishingly well, and the team is working, with Gleason’s input, to make it better and get it into the hands of more people it can help.

Imagine being able to predict epidemics before they occur. Project Premonition is aiming to do just that, with the (involuntary) help of mosquitoes. And drones. The idea is to capture the insects and analyse them to see what pathogens they’re toting around, before they manage to infect populations. That involves determining where the mosquitoes congregate, then placing traps, retrieving them and identifying what bugs the bugs are carrying. Done manually, researchers can only place half a dozen traps in a 14 hour day.

Enter mosquito-as-a-device. Studies have shown that even diseases that don’t infect mosquitoes can be detected through analysis of the insects. Project Premonition concentrates on finding ways to capture the bugs, retrieve them, and use gene sequencing and analytics to detect pathogens.

droneThe design uses drones to map the target area and determine where mosquitoes may be found. Then automated smart mosquito traps are placed. Those traps can sense what kind of insect enters each cell, and only retain mosquitoes. The plan is to have the traps retrieved by drone (something the team is still working on), gene sequence the mosquitoes, drop that information into the cloud and analyse it. Through analysis of volumes of information about microbes and viruses in space and time, researchers hope to see the movement and evolution of possible pathogens before they cause disease in humans, and to do so cheaply enough to make the technology practical for developing countries.

When you have a video involving a lot of motion – say, a ski run, or even a walk down the beach – the smoothest track at normal speed can induce seasickness when it’s sped up, bouncing and jittering no matter how careful you are while shooting it. Microsoft Hyperlapse was developed to make those videos watchable at accelerated speed. Algorithmic magic smooths the bumps and makes the video play well at pretty much any speed. So far, the project has produced a free Android app, and paid software for Windows and Mac OS X, with more to come.

Computers are good at a lot of things, but recognizing and classifying faces and detecting emotions tend to be reserved for super high-powered systems. Even then, they don’t always work very well. Many vendors are struggling with these problems, and Microsoft is no exception. Researchers at Project Oxford are working on software components (APIs) for photo and video analysis, speech analysis, and language recognition. We’ve already seen them in action in fun apps such as Microsoft’s Movember MyMoustache app that recognized and rated facial hair, but there are more practical (and probably lucrative) uses for the technology.

Emotion recognition
Emotion recognition

Emotion detection, for example, could help marketers determine consumer reaction to products and displays. The currently available emotion tool can be used to create systems that recognize eight core emotional states – anger, contempt, fear, disgust, happiness, neutral, sadness or surprise – based on universal facial expressions that reflect those feelings. The facial recognition tool can figure out the gender and approximate age of customers – in theory. Results vary on versions I’ve seen – that’s why this is still a research project!

The Language Understanding Intelligent Service (LUIS) interprets spoken or typed natural language. It uses machine learning to help apps understand what a user wants to do, even if he or she deviates from known commands. For example, to start an exercise app’s activity tracking, the user might say “I want to start my run,” “begin a run” or even “go for a run,” and LUIS would figure out that it means that it should begin tracking the person’s distance, and that the type of activity is a “run”. Similarly, language like “please stop my run,” “let’s pause this run” or “I’m done running” would all signal that the system should stop tracking.

Power Map for Office has graduated from research to the real world, now appearing in Microsoft Excel. Yes, there’s been mapping software for a long time, but this kind of 3D mapping is newer. The demo we saw plotted earthquakes on a map of the globe – including their magnitude and where in the earth’s crust they occurred. Yes, in – you could rotate the image to see how deep the quake was. In Excel, you can map virtually any suitable dataset, such as, for example, traffic patterns.

Image from Worldwide Telescope
Image from Worldwide Telescope

The technology is also in use in the Worldwide Telescope, where you can explore the universe from your computer. Microsoft describes it as an observatory on your desktop, and says it “allows you to see the sky in a way you have never seen it before through individual exploration; multi-wavelength views; stars and planets within context to each other; the ability to zoom in and out; and the capability to create, search and view guided tours of the universe. You can view the entire solar system in 3-D with light and shadows created from the sun, and can explore the Earth and Mars in incredible detail. You also can watch planets orbit around the sun, and moons orbit around planets.”

These are just a few of the unexpected things that Microsoft is working on. Anyone can try them on by visiting The Garage, a home within Microsoft for hackers, makers and people who love to play with tech. And it’s through that kind of play that the technology of the future will evolve.