A Multi Hypothesis Data Association Algorithm Based on Bi-GRU
Manman Hu, Mingjie Qiu, Shengli Wang, Peng Zhang · 2023
In radar multi-target tracking technology, data association is used to solve the problem of matching measurements with targets. Traditional association algorithms can be divided into single hypothesis and multi-hypothesis algorithms. When many echoes fall within the same wave gate, it is difficult for single hypothesis algorithms to correctly discriminate the sources of each measurement. Multi-hypothesis algorithms sacrifice computation for association correctness, making them unsuitable for tracking tasks with high realtime requirements. To address these issues, this paper proposes a novel association algorithm based on the Bidirectional Gated Recurrent Unit (Bi-GRU) framework, and redefines the generation principles of the confirmation matrix. The simulation results demonstrate that this approach, compared to traditional association algorithms, can handle common ambiguities and uncertainties in measurement allocation more effectively, achieve higher association accuracy and smaller filtering error.