Strong stabilized joint probabilistic data association method of autoregulatory gating in vehicle nonlinear multi-target tracking

Abstract

This paper proposed a strong stabilized joint probabilistic data association approach based on numerical stability optimization and autoregulatory gating strategy for multi-target tracking of vehicles with multi-maneuvering states in a nonlinear cluttered environment. The approach utilized an enhanced square-root cubature Kalman filter to achieve the optimization for joint probabilistic data association, which advanced the accuracy and stability of nonlinear state estimation during the multi-target tracking. On this basis, an autoregulation method of association gate based on tracking error sensitivity was proposed to improve the data association gating strategy, which effectively advanced the success rate of target association and tracking accuracy. The virtual simulation traffic scenes of intelligent driving were constructed by PreScan to better reproduce the multi-target tracking problem. The results of the simulations show that the method has clear advantages in terms of target tracking precision, numerical stability and environmental adaptability.