Motion Based Online Calibration for 4D Imaging Radar in Autonomous Driving Applications

This paper proposes a motion-based online extrinsic calibration algorithm for a 4D automotive radar, which can also be applied on a 3D radar for its azimuth alignment and position estimation. The process does not require any calibration jig. The algorithm contains two steps with radar measurements and ego-vehicle's motion parameters as inputs. The first step calculates radar's mounting angles (both elevation and azimuth for a 4D radar), based on which the second step calculates the position of the radar on x- and y-axis of the ego-vehicle. Real world measurements collected by a 3D radar and motion sensors are used for the estimation. Comparing the results with reference values provided by a laser rangefinder shows that the algorithm has a good performance.