Motion Compensation Based Track-before-detect Methods for Automotive Radar System

This paper considers the extension of Dynamic programming based Track-Before-Detect (DP-TBD) to the automotive radar system. DP-TBD is a grid-based method that always be considered in ground-based radar in existing work. For ground-based radars, the surveillance area between different scans is overlapping, and each measurement cell of different scans is matched. However, there is platform motion for automotive radars, which results in misaligned grids. Furthermore, there are also multiple coordinate systems that need to be considered during the implementation of DP-TBD. Consequently, the ground-based DP-TBD radar cannot be directly applied to automotive radar because of the following two aspects, i.e., mismatch of measurement cells and unknown relative motion. To address these problems, this paper proposes a motion compensation strategy built upon the DP-TBD algorithm (MC-DP-TBD). We search for valid state transition space in the navigation coordinate system and integrate the merit functions in the antenna coordinate system, so as to effectively reduce the errors that occur in the previous algorithm. Both simulation data and real radar data are used to demonstrate the superiority of the proposed method in terms of detection performance and estimation accuracy.