As we approach the one year anniversary of the Kinect launch, Microsoft has announced that the Kinect for PC Commercial SDK will be released in early 2012 (http://majornelson.com/2011/10/31/xbox-360-celebrates-one-year-anniversary-of-the-kinect-effect/). More than 200 businesses worldwide, including Toyota, Houghton Mifflin Harcourt and Razorfish, are involved in a pilot program to explore the commercial possibilities of the Kinect.
Until now, most companies working with the Kinect have been working within the constraints of a research license for the Kinect SDK. Consequently the applications that corporations have been working on have been restricted to tightly held private projects or, at most, proof-of-concept projects visible only as demo reels on the Internet. While most people are at least aware of the Kinect technology, the terms of the research license has relegated it to being an afterthought or something only understood at a distance – a nice to have.
The recent announcement of the timeline for the commercial license implicitly green lights these projects to make preparations for releasing Kinect-enabled applications for everyday use. Over the next year we can expect to see the Kinect as a ubiquitous part of our daily environments and something just as prevalent as interactive kiosks are today. The spread of the Kinect beyond the living room may be as dramatic as the proliferation of smart phones or tablets – one day no one knew what they were and, the next, everyone seemed to have one. In boardrooms across America, the question will no longer be one of whether to have a Kinect strategy but instead what that strategy is.
As the Kinect becomes more prevalent in our daily lives, the possibilities and limitations of the Kinect will undergo much closer scrutiny. The potential offered by a mass produced device that provides a video camera, an infrared depth camera and a four microphone array with beamforming capabilities is vast. The technology can be taken in multiple directions including computer vision in robotics, 3D modeling with multiple linked devices, inexpensive augmented reality, hands-free interactive experiences, speech recognition based in-store assistance and innovative computer assisted learning.
That Microsoft’s visionary strategy in designing the Kinect has revolved around off-loading processing to the operating system rather than building it solely into the hardware means that complex scenarios not currently supported by the Xbox can be made viable through improved software and processing power on computers and video cards, the price of which are constantly falling. Microsoft’s Kinect technology is actually scalable and does not require improving the Kinect hardware itself but, instead, on simply improving the software that processes the data streamed by the Kinect.
This all leads to the inevitable question – what is the future of the Kinect? After a year, what are second generation Kinect applications going to look like? The answer depends on where Microsoft takes Kinect software going forward. The current research version of the Kinect SDK beta shows its roots in gaming. The visual processing, depth processing and even acoustical models are tied to the limitations and optimizations required for the Xbox 360 gaming system. They all work best in a room about the size of your living room and even begin to have troubles in small apartments. The microphone array seems to work well in standard rooms, for which it has painstakingly been optimized to deal with surround sound speakers and audio reflections off of furniture, but appears to have trouble in large spaces.
Strikingly, even though the depth camera is capable of 640 x 480 resolutions, the current SDK only provides access to 320 x 240 image streams. The Kinect SDK, likewise, does not provide depth data information for objects within 800 mm (about 2 ½ feet) of the Kinect sensor even though the camera does capture this information.
There are clearly performance reasons for setting these limitations. However part of the problem also appears to be related to the fact that the USB connector for the Kinect is a bottleneck and has been throttled for the particular USB controller configuration requirements of the Xbox. As the Kinect moves out of the living room and into the real world, it makes sense to leave the restrictions imposed by tying the Kinect SDK to the Xbox behind. If we can use improved software running on improved hardware to boost the capabilities of Kinect for PC applications, it would be a shame to have a gaming infrastructure be the main showstopper.
Nowhere is this more clear than when we consider using the Kinect in the office. As a Kinect developer, I have to slide my chair back and away from my monitor whenever I want to debug a piece of code. Fortunately I don’t work in a cubicle and have some open space behind me. I am also fortunate that my chair has wheels and I have the code – slide – code routine down pat. However I don’t see anyone wanting to use a Kinect-enabled business application in this way. Unlike the living room, which is the natural space of our home lives, the office environment of our work lives is generally cramped and close to the screen with just enough room for a keyboard between us and our monitor. We are always within two and a half feet of the objects we work with.
Yet the workspace is one of the chief places we want to see our Kinects working. And instead of large arm movements, we would like to wave our hands or snap our fingers in order to make things happen on our screens. We want The Minority Report writ small. In order to achieve this, in turn, we need to move beyond skeletal tracking and start enabling fine finger tracking.
Along the same lines, for larger movements, the skeletal tracking capabilities of the Kinect only work with the full body. At the office, sitting in our office chairs, we typically never see anything below the waist. Even skeletal tracking, then, needs to be modified to take this into account and to support partial skeleton tracking at the software level.
As the Kinect is being allowed to travel beyond our living rooms with the upcoming release of the commercial Kinect SDK, the software that allows developers to build applications for the Kinect needs to cut its strong dependence on gaming scenarios. This is the natural future for a technology that is maturing. This is where the Kinect is headed – not only out into the world but also up in our faces. We want and need to get closer to the Kinect.