Title: Extrinsic Calibration of 2D mm-Wavelength Radar Pairs Using Ego-Velocity Estimates

Abstract: Correct radar data fusion depends on knowledge of the spatial transform between sensor pairs. Current methods for determining this transform operate by aligning identifiable features in different radar scans, or by relying on measurements from another, more accurate sensor (e.g., a lidar unit). Feature-based alignment requires the sensors to have overlapping fields of view or necessitates the construction of an environment map. Several existing methods require bespoke retroreflective radar targets. These requirements limit both where and how calibration can be performed. In this paper, we take a different approach: instead of attempting to track targets or features, which can be difficult in noisy radar data, we instead rely on ego-velocity estimates from each radar to perform calibration. Our method enables calibration of a subset of the transform parameters, including the yaw and axis of translation between the radar pair, without the need for a shared field of view or for specialized structures in the environment. In general, the yaw and axis of translation are the most important parameters for data fusion, the most likely to vary over time, and the most difficult to calibrate manually. We formulate calibration as a batch optimization problem, prove that the radar-radar system is identifiable, and specify the platform excitation requirements. Through simulations studies and real-world experiments, we establish that our method is more reliable and accurate at estimating the yaw and translation axis than state-of-the-art methods. Finally, we show that the full rigid-body transform can be recovered if relatively coarse information about the rotation rate is available.