Research on effectiveness of active braking strategy of autonomous vehicles for VRUs in mixed conditions

Pedestrians and two-wheeled cyclists are referred to as vulnerable road users (VRUs). The active braking system can prevent collisions between vehicles and VRUs. Currently, the research on the active braking strategy mainly focuses on the safety and comfort when VRUs laterally cross the straight road with uniform motion. However, considering the variety of roads and the diverse motion states of VRUs, it is essential to explore the effectiveness of the active braking strategy in mixed conditions where VRUs diagonally and laterally cross the curved road with different motion trajectories and speeds. Firstly, the location relationships between the vehicles and VRUs are determined to establish the prediction model of VRUs’ motion state and the safety evaluation model. Secondly, based on the linear quadratic regulator and supervised Hebb learning rule, the collision avoidance controller is devised. Finally, the proposed active braking strategy is verified through the joint simulation platform and hardware-in-loop tests. The results show all crashes between vehicles and electric bicycles can be avoided. The braking strengths range from 0.35 to 0.71, the braking durations range from 2.34 s to 3.97 s, and the peak braking pressures are less than 75 bar, which can guarantee the comfort of occupants.