State Parameter Estimation of Intelligent Vehicles Based on an Adaptive Unscented Kalman Filter

The premise of vehicle intelligent decision making is to obtain motion state parameters accurately and in real-time. Several cannot be measured directly by sensors, so estimation algorithms based on filtering are effective solutions. most representative algorithm the Kalman filter, especially standard unscented filter (UKF) that has been widely used because its superiority dealing with nonlinear problems. However, although UKF assumes noise statistics system known, due complex changeable operating conditions, sensor aging other factors, these noises vary. In order realize high-precision estimation, a noise-adaptive proposed this article. maximum posteriori (MAP) dynamically update system, it embedded into step form an adaptive (AUKF). will when unknown prevent divergence adjusting mean covariance estimated improve accuracy. On basis, method verified joint simulation CarSim Matlab/Simulink, confirming AUKF performs better than accuracy stability under different degrees disturbance, for yaw rate, side slip angle longitudinal velocity improved 20.08%, 40.98% 89.91%, respectively.