Abstract:

Aiming at the problem of large tracking error caused by the change of target motion state in the process of tracking complex maneuvering targets, an adaptive interacting multiple model unscented kalman filter （IMM-UKF） algorithm is proposed, which uses model probability posterior information and model likelihood function to adaptively modify Markov transition probability matrix （TPM）. The model probability correction method and model transition acceleration method are designed. The two methods act on the model stability stage and model transition stage respectively, which improve the model probability accuracy and model transition response speed, and reduce the state estimation error. Finally, the performance of the proposed algorithm in the target with complex motion is verified by experiments in two scenarios, and compared with the traditional methods, the position accuracy and velocity accuracy are improved by 15% and 26% respectively when the target is maneuvering, which verifies the effectiveness and feasibility of the algorithm.

Key words: target tracking, interacting multiple model （IMM）, adaptive, unscented Kalman filter （UKF）