Abstract:

When tracking multiple targets with radar, traditional tracking algorithms are typically implemented based on single-frame recursive processing. As these methods only consider single-frame information from multi-target echoes, they tend to accumulate significant errors. To comprehensively utilize multi-frame target information, this paper proposes a multi-objective adaptive trajectory-oriented estimation algorithm based on trajectory random finite sets. The method first generates multiple candidate measurement sequences within the observation area using trajectory random sets, then performs polynomial time series analysis on these measurement sequences to achieve holistic trajectory estimation, thereby directly generating complete motion trajectories for multiple targets. The trajectory random set approach incorporates multi-frame motion information through trajectory hypothesis modeling, significantly reducing the impact of clutter environments on multi-target tracking performance. Meanwhile, the holistic trajectory estimation method directly estimates the complete motion characteristics of targets. Its adaptive model can dynamically adjust with incoming observation data, enabling continuous-time state estimation at arbitrary moments within the observation interval. Simulation results demonstrate that the proposed algorithm achieves effective multi-target tracking in cluttered environments. Compared with conventional algorithms, this approach generates smoother complete trajectories and exhibits superior multi-target tracking performance.

Key words: multiple target tracking, trajectory-oriented estimation, trajectory random finite set, polynomial time series analysis