Micro-Drone Ego-Velocity and Height Estimation in GPS-Denied Environments Using an FMCW MIMO Radar

Abstract

In the context of autonomous navigation, the vehicle trajectory estimation and the detection of surrounding obstacles are two critical functionalities that must be robust to difficult environmental conditions (e.g. fog, dust, snow) and the unavailability of infrastructure signals (e.g. GPS). With the advantage of remaining operable in low-visibility conditions, radar sensors are good candidates to detect obstacles in an autonomous navigation context. In this paper, we show that radars can also be successfully used for real-time trajectory estimation. We address the case of an autonomous micro-drone intended for the exploration of piping networks and embedding a Frequency Modulated Continuous Waves (FMCW) MIMO radar. We show that using a beamforming technique to virtually steer the radar field-of-view, we can simultaneously estimate the horizontal and vertical velocity of the drone as well as its height. These results are first validated through simulations based on experimental drone flight data and a radar simulator. Then, using an Infineon 77GHz FMCW radar, we show through real-world experiments the high performance attainable with our solution.