From a smartphone app capable of capturing detailed 3-D scans to interactive whiteboards to a program capable of visualizing in several ways how a social media message goes viral, Microsoft’s TechFest 2013 was certainly full of cool stuff.
TechFest is the company’s annual science fair in which its advanced researchers show demos of what they’re working on. Today, a small portion of the approximately 150 demos are being shown to media, customers and partners. Wednesday and Thursday, TechFest will be open to Microsoft employees.
More than just cool stuff, though, the demos — or at least the small portion shown in the preview today — brought into focus some research areas Microsoft has been working on for several years now: natural user interface (NUI) — meaning interacting with computing devices using touch, speech or gestures; big data — synthesizing and making useful large amounts of information; and machine learning — the ability of computers to learn.
What came through at this TechFest is how those three areas often work together and also how “all these technologies are coming to maturity — both at Microsoft and in the industry at large,” said Steve Clayton, who writes about Microsoft Research for the company.
One example of that that Clayton offered: The keyboard on Windows Phones uses big data and machine learning to make the touch user interaction better. The Windows Phone keyboard is able to predict what you’re about to type based on what people have searched for in Bing and, over time, on what you tend to type. The keyboard responds by displaying in subtly larger size the next key for the word it believes you’re about to type.
The goal, at least for Microsoft Principal Researcher Bill Buxton, is to get to the point where these technologies come together and become transparent — when you don’t even think of your computing device as one.
For instance, a smartphone, when connected to a car, becomes part of the car, allowing you to do different things than if you hooked the smartphone up to a large digital display at your workplace. Where you use your phone then “fundamentally changes the nature of your phone” so that your smartphone and the screen come “seamlessly together so it’s one device,” Buxton said. “If you think ‘computer,’ it’s a failure of design.”
Here are some demos I saw today in which those themes came into play:
Kinect Fusion (above) enables 3-D scanning using a Kinect sensor. The goal is to make 3-D scanning easy and affordable, said Shahram Izadi, a Microsoft researcher. The new version of Kinect Fusion that he demonstrated will be part of the next Kinect for Windows SDK release, he said.
SketchInsight (above) allows people to visualize data using simple sketching gestures. For instance, once a set of data has been loaded, a person could start sketching everything from a simple “L” shape to a circle on a digital whiteboard and, using the data on hand, SketchInsight quickly turns that into a graph chart or a pie chart.
Kinect hand recognition (above): Currently, the Kinect motion sensor reads large, skeletal body motions — think of it as sensing a stick figure in motion. Now researcher Cem Keskin is training the Kinect to read more detailed hand gestures. An open palm held toward the sensor, for instance, moves a cursor on a large display screen. A closed fist allows the user to “grip” things on that screen — similar to what clicking-and-dragging on a mouse does. Moving two gripped fists closer or father apart allows you to zoom in or out of the screen. These capabilities allow people to use Windows touch gestures without having to touch the screen. It will be in the next version of Kinect for Windows SDK, Keskin said.
SandDance (above) allows users to visualize large amounts of data both on a grand level and on a very granular — sand-like — level. Data — for example, counties in the U.S. where more than 60 percent of people voted for Obama in the last presidential election — can be displayed as grains of sand. Tapping on each grain can pull up detailed information about that county.
I’m at the Microsoft campus in Redmond for TechFest, the annual event where the company’s advanced researchers give demonstrations of the cutting-edge research they’re working on.
Microsoft employs some 850 Ph.D-level researchers worldwide — about half in the U.S. — and spends about $9 billion a year on research and development.
That makes the company the No. 1 organization for computer science research, according to Rick Rashid, Microsoft’s chief research officer, in his keynote address this morning.
There have been questions about whether Microsoft gets good return on its heavy R&D investment.
Rashid addressed some of that during his keynote, saying that Microsoft Research (MSR) generates about a third of the company’s patents (a slide he was showing said MSR generates about 25 percent), that it brings smart people to work on hard problems confronting products or the company and that it provides “early warning” on new technologies and allows companies to respond more rapidly to change. MSR’s work also ends up in almost all products that Microsoft ships, he pointed out.
As an example of projects that MSR is working on that might have application to Microsoft products, Microsoft researcher Curtis Wong offered a preview glimpse of GeoFlow, which uses technology behind the WorldWide Telescope (developed by Microsoft Research) to help visualize data in Excel programs. He demonstrated with data from the city of Chicago, allowing the user to zoom in and out of a map to get various information — pot arrests by neighborhood and over time, for instance.
A portion of the many TechFest demos that will be shown later this week to Microsoft employees are being previewed today. The focus this year is on natural user interface (NUI), machine learning and big data. I’ll be visiting some of the demos and will update this post with some highlights.