Automated Braking Decision and Control for Pedestrian Collision Avoidance Based on Risk Assessment
Lin Zhang,Jiawei Yuan,Kang Yuan,Jinlong Hong,Haitao Ding,Hong Chen
DOI: https://doi.org/10.1109/mits.2021.3098618
IF: 5.293
2021-01-01
IEEE Intelligent Transportation Systems Magazine
Abstract:An automated braking decision and control framework is proposed in this article to avoid intelligent vehicle–pedestrian collisions. The framework contains three parts: pedestrian collision risk assessment, automated braking decision making, and automated braking control. First, a kinematic vehicle model and a Kalman filter are used to estimate the path trajectories of a host vehicle and a pedestrian, respectively, by using onboard sensor information. Then, pedestrians with the highest risk levels are selected as the most dangerous collision objects from the valid target pedestrian sets. Second, based on road adhesion conditions, vehicle lateral stability conditions, and driver intention constraints, an automated braking decision strategy is proposed using fuzzy theory; in this approach, the collision time difference is the input, and the desired deceleration is the output. Third, a first-order, inertial pure hysteresis braking system response model is built based on real vehicle braking experimental data. The relationship between braking deceleration and the master cylinder pressure is established through data fitting to achieve automated braking control for the host vehicle. Finally, experiment results indicate that the proposed control framework can calculate reasonable deceleration rates and perform braking control to achieve complete collision avoidance, thus verifying the effectiveness of the algorithm.