Correction adaptive square-root cubature Kalman filter with application to autonomous vehicle target tracking

For a state estimation problem of nonlinear system, the square-root cubature Kalman filter (SCKF) is an effective method when the noise statistical characteristics are known. However, the performance of SCKF often degrade significantly in the face of uncertain noises interference, particularly in case of measurement or system failure. In this paper, we focus on improving the accuracy and robustness of SCKF under irregular noise. First, a weighted adaptive SCKF (WASCKF) algorithm is presented with moving window method. The WASCKF can improve the accuracy of SCKF by adaptively adjusting the covariances of measurement noise and process noise. Next, in order to further improve the robustness of WASCKF against the abrupt abnormal noise, a correction adaptive SCKF (CASCKF) algorithm based on fault detection mechanism is proposed. The CASCKF algorithm can detect whether there is a fault according to a statistical function of Chi-square distribution, and can judge and carry out the necessary correction processing by using an isolate rule. Finally, the performance of CASCKF is verified by numerical experiments of autonomous vehicle target tracking problem. The results show that the proposed CASCKF algorithm has good accuracy and robustness even with sudden abnormal noise interference.

Automated summary

In this paper, a weighted adaptive SCKF (WASCKF) algorithm and a correction adaptive SCKF (CASCKF) algorithm are proposed to improve the accuracy and robustness of SCKF under irregular noise. WASCKF enhances accuracy by adaptively adjusting noise covariances, while CASCKF improves robustness by using fault detection mechanism and isolate rule. The numerical experiments on autonomous vehicle target tracking problem demonstrate that CASCKF algorithm has good accuracy and robustness against sudden abnormal noise interference.