Research On Vehicle Radar Target Tracking Algorithm Based On DSP

In this paper,target tracking algorithms for automotive radar is studied and a target tracking system for automotive radar is implemented in DSP embedded system.Firstly,a maneuvering model based on support vector regression(SVR)is proposed to overcome the large errors in the application of a priori maneuver model,such as the Singer model and the CS model,in the application of vehicle radar tracking to strong maneuvering targets.This model takes the difference between the theoretical innovation covariance and the actual innovation covariance as the criteria for judging the target maneuvering,and adjust the model parameters online by the support vector regression algorithm.A large number of experiments show that the proposed model based on support vector regression is superior to the CA model and the Singer model in tracking the strong maneuvering target.Secondly,fast square root cubature kalman filter(FSRCKF)is put forward to improve the square root cubature kalman filter(SRCKF),which gets poor real-time perfomance in embedded system.FSRCKF combines the time update of linear Kalman filter and measurement update of SRCKF.A lot of simulation experiments show that the improved FSRCKF has the same filtering precision compared with SRKCF,while filting time consumption is reduced by 30%or more,which indicates that FSRCKF is much more suitable for embedded systems with high real-time requirements but lack of computing resources,such as the automotive radar target tracking system.Finally,an automotive radar target tracking system is realized based on DSP embedded environment and practical performance of partial motion model and filter is verified.Numerous experiments show that the system has high usability and meets the requirements of system design.