Instantaneous Ego-Motion Estimation Using a Coherent Radar Network

This paper presents an approach to estimate the ego-motion of a vehicle equipped with a network of two or more coherently working bistatic or multistatic MIMO radars. The multi-perspective view of this setup allows for an immediate determination of the radar targets' relative velocity vector. The ego-motion is then derived from the inverted vectorial velocity of all static targets that can be obtained by clustering the velocity vectors. For this, it can be assumed that the majority of the detected targets are static, thus generating the largest cluster. The novel approach was tested with a setup consisting of two 77 GHz front-looking, long-range radars that only exhibit a narrow field of view with a small azimuth variation. The functionality of the proposed algorithm was evaluated and confirmed in a highway test drive.