Abstract: For the problem of low recognition accuracy for the target vehicle of single sensor, this paper proposes a target tracking method based on millimeter-wave radar and camera information fusion. Meanwhile, based on extended information fusion (EIF) and unscented information fusion (UIF), checking unscented information fusion (CUIF) algorithm is established to fuse multi-sensor information. The CUIF algorithm uses a distributed fusion structure to correlate and fuse the trajectories from different sensors to obtain the target trajectory, and applies the distributed detection system to perform real-time system state compensation on the acquired sensor delay information, thus solving the problems of single sensor signal delay and packet loss. CarSim and Simulink are used jointly to build a simulation and test platform to validate the algorithm of the established fusion model. The experimental results show that CUIF algorithm reduces the target loss rate by more than 10% compared with EIF and UIF, and reduces the time delay to 5 ms, which meets the real-time requirements of the system.

Key words: information fusion, checking unscented information fusion algorithm, distributed fusion structure, distributed detection system