Joint Velocity Ambiguity Resolution and Ego-Motion Estimation Method for mmWave Radar

In this letter, a joint velocity ambiguity resolution and ego-motion estimation radar odometry method is proposed to solve the radar odometry algorithm invalidation caused by radar velocity ambiguity. First, a signal model for the radar odometry is developed for vehicle ego-motion estimation in the presence of velocity ambiguity. Then, a guided radar angle-velocity point cloud clustering algorithm is designed and the solution of the established signal model is presented. The estimation performance of radar odometry before and after velocity ambiguity resolution is quantitatively analyzed, which provides ideas for improving the estimation performance of the radar odometry. Finally, the radar-based vehicle pose estimation algorithms were compared in the presence of velocity ambiguity in real-world road environments. By applying the proposed algorithm to radar SLAM, a high-precision localization with an average translation error of less than 0.28 meters and an average rotation error of less than 0.97 degrees was achieved in a circular road track with a total mileage of more than 500 meters.