Interferometric Radar For Spatially-Persistent Gesture Recognition in Human-Computer Interaction
Eric Klinefelter, Jeffrey A. Nanzer · 2019
We present a new method of spatially-persistent radar hand gesture recognition based on direct measurements of the angular velocity of hand motions for improved-human computer interaction. Traditional radar-based gesture recognition utilizes Doppler-based measurements, however the radial velocity and thus the Doppler time-frequency responses change significantly when the hand is not directly broadside to the radar, leading to degraded classification of gestures. Using a newly developed interferometric method of directly measuring angular velocity, we show that the time-frequency signatures of hand gestures remain persistent when moving away from broadside, indicating that the interferometric technique can be used for more robust and reliable wireless human-computer interaction. Experimental measurements using a 16.9 GHz continuous-wave interferometric radar demonstrate that for the Doppler mode the maximum response bandwidth degrades by 71% when the gesture moves to 30° off-broadside, leaving a negligible time-frequency response, while the interferometric signature generated from the same data degrades in maximum bandwidth by only 16%, with the response remaining largely identical.