An Efficient TBD Method for Automotive Radars with Time- Varying Field of View

This paper explores the detection and tracking of weak targets in automotive radar systems using the multi-frame track-before-detect (MF - TBD) method. The MF - TBD technique has been widely studied in ground-based radar systems. However, an extension of MF-TBD to automotive moving-platform radar systems always faces some challenges, such as time-varying radar field of view (FOV), nonlinear conversion in mixed coordinates, and high computational complexity in implementation. To solve the above problems, we proposed an efficient automotive radar MF - TBD algorithm. The proposed algorithm designs accurate target search paths among multi-frame measurements to avoid performance loss due to the nonlinear conversion in multiple coordinates. Besides, an adaptive track extraction strategy is proposed to solve the FOV misalignment in moving-platform radar systems. Finally, by combining with the low-threshold preprocessing operation, the proposed algorithm strikes a better balance between target detection performance and computational complexity in automotive radar systems. The numerical results indicate that the proposed algorithm delivers improved detection and tracking performance while maintaining relatively low computational complexity.