Estimation of Ego-Position Using Frequency-Modulated Continuous Wave Radar Sensor Data

摘要

To map the surrounding environment using a radar sensor, both radar detection results and robot’s motion information are required. In general, an additional inertial sensor is required to acquire motion information of the robot. In this study, a method for estimating a robot’s ego position from radar detection results without the aid of other sensors is developed. First, the absolute positions of the detected targets are determined by comparing the radar detection results in two consecutive frames. Next, the position of the robot in the new frame is estimated based on the positions of the identified targets. Finally, the position of the robot obtained from the inertial sensor is compared with the position estimated by the proposed method, and the average position estimation error is found to be less than 0.33 m.