Resource Efficient Gesture Sensing Based on FMCW Radar using Spiking Neural Networks

Gesture recognition is a natural and intuitive human computer-interface compared to traditional interfaces such as mouse and keyboards. Radar forms a promising modality for portable gesture recognition systems, where minute finger motions can be easily sensed and processed to extract meaningful information presenting the user's intention. Compared to several conventional deep learning approaches that have been proposed in the literature, in this paper we present a novel spiking neural network (SNNs) for gesture recognition using a 60-GHz frequency modulated continuous wave radar (FMCW) chipset. SNNs are more hardware friendly and energy-efficient than their deep learning counterparts making them attractive for portable devices. We have demonstrated in measurement the application of SNN for the processing of radar data.