Abstract: In order to improve the multi-target tracking accuracy index of vehicle-mounted millimeter-wave radar and enhance the driving safety of road vehicles, based on the fusion of interactive multi-model unscented Kalman filter (IMM-UKF) and joint probabilistic data association (JPDA), according to improved strong tracking UKF (ISTUKF) an IMM-JPDA-ISTUKF algorithm is proposed in this paper to solve the problems of poor robustness and low filtering accuracy of UKF at abrupt changes in vehicle motion state. The performance of the algorithm is verified by simulating road scenes and building a simulation environment. In order to prove the improvement of the tracking accuracy of the algorithm under actual road conditions, the radar road test is also carried out, and the effectiveness of the algorithm is further verified by the vehicle data obtained by the radar on the road. The results show that the algorithm improves both distance tracking accuracy and speed tracking accuracy when the motion states of the target vehicle are changing.