An Integrated Approach for Vehicle State Estimation Under Non-Ideal Conditions Using Adaptive Strong Tracking Maximum Correntropy Criterion EKF

Accurate acquisition of critical vehicle states is a prerequisite for active safety systems to work properly. However, vehicle sates under non-ideal conditions are usually difficult to be measured directly due to the high cost of sensors. To deal with the problem, an adaptive strong tracking maximum correntropy criterion extended Kalman filter (ASTMCC-EKF) is put forward to estimate vehicle states. Maximum correntropy criterion (MCC) is introduced as the optimization criterion to construct the cost function. Strong tracking filter is employed to dynamically adjust the prior error covariance matrix. Moreover, Sage-Husa suboptimal unbiased estimator is adopted to estimate the noise in real time. Simulation experiments and road tests show that ASTMCC-EKF has a more excellent estimation performance than existing algorithms under non-ideal conditions. The algorithm not only effectively suppresses the interference of non-Gaussian noise but also improves the robustness of ASTMCC-EKF under the model uncertainty and time-varying noise. Furthermore, the proposed ASTMCC-EKF shows a strong robustness to different driving conditions and road conditions.