Event-triggered shared lateral control for safe-maneuver of intelligent vehicles

• Published in: Science China. Information sciences 2021-07, Vol.64 (7), p.172203, Article 172203
• Main Authors: Jiang, Yan, Zhang, Xinglong, Xu, Xin, Zhou, Xing, Dong, Zhengzheng
• Format: Article
• Language: English
• Subjects: Active control; Artificial neural networks; Automobiles; Collision avoidance; Computer Science; Control algorithms; Control systems; Controllers; Cooperation; Driving; Fines & penalties; Human factors; Information Systems and Communication Service; Intelligent vehicles; Lane keeping; Lateral control; Linear quadratic regulator; Maneuvers; Neural networks; Optimization; Predictive control; Research Paper; Risk assessment; Risk levels; Simulation; Steering; Traffic accidents; Traffic accidents & safety; Vehicles
• ISSN: 1674-733X; 1869-1919
• DOI: 10.1007/s11432-020-2961-8

With the increasing number of cars, road traffic accidents have caused a lot of losses every year and human factors play an important role in many cases. Applying active safety assistance control or shared control techniques in intelligent vehicles is promising to reduce the number of traffic accidents. In this context, the dynamic optimization of the shared control policy and the smooth transitions of control authority between human drivers and intelligent driving systems are critical issues to be solved. Motivated by this, this paper proposes an event-triggered shared control approach for safe-maneuver of intelligent vehicles with online risk assessment. In the proposed approach, a Bayesian regularized artificial neural network (BRANN) is designed to predict vehicle trajectories and build a quantization function to assess the risk level owing to potential collision events. The shared controller dynamically optimizes the shared control policies between the human and the intelligent driving system via solving a model predictive control (MPC) problem. The predicted driving behaviors in the prediction horizon are pre-computed with a finite-horizon model predictor steering the predicted trajectories contributed by human driving. Moreover, smooth transitions back to human driving mode are realized via adding penalties on the shared control of the intelligent driving system. Three simulation scenarios in the PreScan environment, i.e., rear-end collision avoidance, lane-keeping and unskilled driving, are studied to test the effectiveness of the proposed approach. The simulation results, including the comparison with a linear quadratic regulator (LQR)-based shared controller, are reported, which show that the proposed approach can timely evaluate dangerous events and realize safe driving in terms of collision avoidance and lane-keeping. Also, the proposed approach outperforms the LQR-based shared controller in terms of smooth transitions.