Microsoft Research and the University of Washington have developed Kinect-like gesture control software that can be easily implemented into almost any existing computer system. The technology is called SoundWave – a program that utilizes standard speakers and microphones to enable motion control in existing PC setups.
SoundWave operates using the Doppler effect, where an inaudible frequency (18-22 KHz) is sent out through a computer’s speakers and altered as it bounces off the subject’s hand moving in front of the computer. This altered frequency is read back by the SoundWave program through the computer’s microphone, allowing the system to recognize and carry out corresponding actions based on programmed hand movements. The project’s research paper states that SoundWave can take into account the target’s speed, direction, proximity, and size, all based off alterations of the given frequency.
The video below showcases some of the ways the Microsoft Research team are utilizing the technology, including page scrolling, Tetris playing, and user proximity recognition.
According to the team behind the technology, silence is not a necessity in order for the emitted frequency to be read. Researchers were able to play music and even use the program in a crowded cafe and still maintain the same level of accuracy as quiet conditions.
The most important feature of SoundWave is that it allows gesture recognition to be implemented into nearly any existing computer system, which would otherwise have to undergo extensive hardware augmentation in order to recognize gesture inputs. The team tested it on eleven different computers, including both desktops and laptops, with six different test subjects to ensure effectiveness.
There were some inevitable drawbacks to the SoundWave technology noted in the team’s paper, admitting that while the high frequency tone will be inaudible to most adults, it can potentially annoy animals and children. The paper additionally recognizes the limitations of the Doppler effect when detecting more complex gestures and static poses, though researchers are hoping that an array of microphones will allow for more accurate and complex gesture recognition in the future.
For now, the Doppler-based system may not be as accurate as Kinect’s 3D scanning technology, but it’s likely to be a whole lot cheaper. Be sure to check out the team’s paper to get all the intricate, hand-wavy details.