AUKF-Based Active Collision Avoidance of Vehicles Considering Measurement Noise Uncertainty

Advanced driver assistance systems, especially active collision avoidance systems, can improve vehicle safety planning paths to avoid obstacles in dangerous situations. Active collision avoidance systems usually decide whether to perform path planning based the gap and velocity in relation to the vehicle ahead, and these signals can be obtained through radar. In previous studies, it was assumed that the measurement noise of the sensors is known and the radar information is simply processed and directly input to the collision mitigation system. However, the uncertainty of the measurement noise may reduce the measurement accuracy and lead to premature lane changes, which reduce traffic efficiency, or rear-end accidents due to late lane changes. To solve these problems, we propose an AUKF method for state estimation to obtain the accurate relative distance and speed of the preceding vehicle. Founded on this information, a driving decision-making method considering tire road friction coefficient is established to decide whether to perform lane-changing path planning or not. A seventh-order polynomial-based path planning algorithm is designed to direct the automobile to achieve active collision prevention by obtaining the desired wheel angle derived from inverse dynamics. Simulink-Prescan co-simulation outcomes indicate that the AUKF can enhance the measurement accuracy of the sensor, as well as the vehicle can avoid obstacles more reasonably under emergency conditions.