Fusion of stereo camera and MIMO-FMCW radar for pedestrian tracking in indoor environments

In this paper, we present a fusion of stereo camera and radar targets to significantly improve the tracking of pedestrians indoors especially suited for surveillance and security applications. Indoor environments pose bigger challenges for pedestrian tracking compared to outdoor environments. This makes pedestrian tracking with only a camera or radar difficult. In this work, we demonstrate that in some preliminary experiments our fusion system increases the multiple object tracking accuracy (MOTA) from -16.7% and 38.0% for camera or radar only tracking to 90.9% using a fusion of both. Furthermore, the proposed fusion system achieves a multiple object tracking precision (MOTP
_3D
) of 90.3% compared to 83.4% and 81.8% obtained via camera and radar only tracking, respectively.