Ubi Interactive Bets on Kinect, Finds a New Home in Seattle
Last Christmas was not shaping up to be a banner holiday season for Anup Chathoth. Sick and cooped up at home, the young entrepreneur was contemplating how he could revive an idea that he’d been crunching for several months.
Chathoth and his partner, Chao Zhang, were trying to make a new device that could turn a wall, table, or virtually any other surface into a kind of touchscreen. By combining a projector, sensitive cameras, and motion-tracking software, they figured it would be possible to make formerly static images into a fully interactive display, with the motion of a user’s fingers standing in for a standard mouse.
Touch the image of a button being projected onto a wall, and their software would interpret it as a mouse click, opening up the next page of a website. Pinch two fingers together on a map, and it would zoom in, just like on a tablet or smartphone.
But when it came to assembling the hardware, Chathoth and Zhang were just plain stuck. Their startup, ubi interactive, had tried to put together its own small projector unit that could be worn around a user’s neck or set on a table. They’d looked into making a peripheral device, something that could be plugged into a computer’s USB port to track movements on an external screen.
Neither option really worked. For one thing, the high-end cameras they were counting on to track the movements were prohibitively expensive. And even in this age of cheap global manufacturing, where entire cities in China are devoted to churning out components and devices, it’s still no small feat to build a new electronic gadget.
“We’re not Apple or Amazon or Google,” Chathoth says. “We’re talking about a device which has a computer, cameras, and wireless—everything.”
That’s when Chathoth remembered Microsoft’s Kinect motion-sensor. Chathoth and Zhang had played with the thing before, but considered its hardware a little too limited for the high-test uses they were planning.
But what the hell, he figured. He was already stuck at home with two weeks of holiday vacation to spare and a nagging illness. And it’s easier to write software than to try making your own hardware.
So Chathoth ran back to Munich’s Technical University, where he and Zhang had been working on their project, and grabbed the Kinect system. And for the next couple of weeks, he coded away.
“By the time we came back from vacation, I was actually setting it up in our room,” Chathoth says. “Chao was coming in and saying `Hey, this works now.’ So that was a truly big moment for us, because suddenly, a lot of our problems disappeared.”
When the Kinect debuted in late 2010, it revolutionized the console gaming industry.
Nintendo had previously made huge waves with its Wii system, which tracked how a player wielded their handled remote control. But the Kinect’s sleek little black bar blew past the Wii and Sony’s PlayStation3, allowing players to control the onscreen action simply by contorting their bodies.
Hackers immediately saw much bigger potential. A wave of customizations began cropping up almost at once, contemplating everything from controlling toy helicopters to surgical robot interfaces. Suddenly, the potential of Kinect was growing well beyond powering video-game dance competitions.
Microsoft’s initial public response was not welcoming, with hints of possible legal consequences for tampering with the company’s new toy. But the tune coming from Redmond soon changed. People working directly on the Kinect project said they’d meant to expose the device to developers all along—the company PR machinery just hadn’t been clued in yet. Some didn’t buy that line, but it hardly mattered. Microsoft was embracing outside alterations of its video-game sensor as a new computer interface.
It was an important step for the company. Xbox is Microsoft’s only surefire consumer device success, winning where the Zune music player, Kin phone, and early versions of tablets have all failed. Getting talented outsiders to work on the platform would only make the success sweeter because it could help advance the uses for Kinect’s motion- and sound-sensing tools far beyond gaming.
Suddenly, the stodgy old enterprise software behemoth that couldn’t innovate to save its life had something going.
Chathoth and Zhang had been watching the growth of Kinect modifications from their post in Munich. At first, Microsoft only offered a Windows development kit for academics and nonprofits. And that version was too limited for what ubi was hoping to accomplish, Chathoth says, because it didn’t expose deep enough data from the device’s processors.
Microsoft promised that a commercial software development kit was on its way. But PrimeSense, an Israeli company that provided Kinect’s chipset, gave developers like Chathoth another opening by releasing its own SDK.
“And there, we get access to all the data. So now, technically, it was possible to build our product,” Chathoth says.
Meanwhile, Microsoft had been getting much more serious about cultivating outside developers for the Kinect program. In November, it announced a deal with big-name startup accelerator program TechStars that would bring 11 startups to Seattle for a three-month product development bootcamp.
Like any big tech company, Microsoft already had initiatives to cultivate startups. But pairing directly with TechStars was a much more innovative approach.
Companies that made the cut would get the standard TechStars investment terms: $20,000 in exchange for 6 percent equity. Microsoft would supply offices, hardware, software, support, and some big-named mentors to the program, but wasn’t taking an equity stake in the companies.
Back in Munich, ubi interactive was intrigued. The small team had put up a test version of its hand-tracking system in the university, powering an information display system that a professor had suggested as a possible use.
Now, hearing about the accelerator—and the upcoming release of an official Kinect SDK for Windows—they thought it would be worth a shot. But time was running short.
“We had like one month to make up our mind and apply for it,” Chathoth says. “So we were making videos and running like crazy to finish our software, to have the video working and everything. And then we applied for it.”
More than 500 teams from around the world did the same. Ubi made the initial cut, and after some additional probing to make sure their product seemed real enough, the startup got the word that it was in the final group of contenders .
There was, however, one last hurdle. Dave Malcolm, a TechStars mentor and former Microsoftie who was directing the Kinect Accelerator, wanted to meet the teams in person before making the final cut. Malcolm got as far as Paris and got ahold of the ubi team in Munich. Could they come meet him?
“So one night, overnight, we were flying to Paris. Monday morning, and we’re in his hotel room. And the first thing he asked was, `It all sounds cool, you know. The problem is, I don’t believe actually that it could work,’” Chathoth says. There was pretty good reason for skepticism. While the Kinect’s skeletal tracking capabilities were well-known, getting the device to track down to the fingertip level was much more complex. Being able to tell when that fingertip was touching a wall, rather than hovering half an inch away from the surface, was an even finer distinction.
Microsoft itself has shown off something in a similar vein before, but that was more of a research project—not close to a product, like ubi was claiming.
The only thing to do was cook up a demo. They put the gear together, dragged a table over to set it all up, and aimed it at the hotel room door. And just like that, Malcolm was playing the touchscreen game “Angry Birds” on his door.
“He immediately said, `You know what? I really want you guys to come to Seattle,” Chathoth says. “So, essentially, we packed our bags and left in the next few days.”
These days, ubi is one of the few teams still in the Microsoft building in South Lake Union, occupying a few desks amid a mass of recently cleared-out cubicles and leftovers from the TechStars program. When we spoke for this story, Chathoth was actually the only member of the company still in town—CTO Zhang and CFO David Hajizadeh have gone back to Munich to square away their affairs before returning to Seattle full-time.
Don’t let the humble surroundings fool you. Ubi interactive seems to be gaining traction already, with some paying customers who want to use the hand-tracking software for corporate conference rooms.
The startup also is eyeing a possible second market in what is called “digital signage,” such as advertising displays for retailers, and is raising a $650,000 round of financing to help it expand the team to include more customer support for its earliest users.
One thing’s clear: Even after an intense accelerator program, and with the sun finally out in the Northwest, this isn’t the time for a summer vacation.
“There is no one who is doing exactly what we do right now. But it’s not going to be the same situation, I’m sure, after a few months,” Chathoth says. “We have an edge right now. To keep it going, we have to move fast.”