关键词: 自动驾驶, 前车侧翻, 轨迹预测, 轨迹规划

Abstract: Addressing the insufficient consideration of uncertain surrounding vehicle motions and the potential instability of self-vehicle control in preceding vehicle rollover scenarios on highways, a multi-constrained high-speed autonomous collision avoidance method is proposed. Firstly, a motion model accounting for uncertainties is employed to predict the trajectories of surrounding vehicles. Subsequently, based on the driving feature vector approach, a comprehensive consideration of various constraints is applied to enhance high-speed collision avoidance safety. This includes the design of four constraint conditions, such as collision avoidance constraints and self-vehicle stability constraints, coupled with four objective functions to optimize and derive the optimal collision avoidance trajectory. Additionally, emergency control strategies are provided in cases where collision avoidance is impossible. Finally, hardware-in-the-loop experiments are conducted in both straight and curved road scenarios. The results demonstrate that the proposed autonomous collision avoidance method effectively ensures safety during high-speed avoidance processes and can initiate emergency braking to mitigate collision risks in collision unavoidable scenarios while meeting real-time requirements.

Key words: autonomous driving, rollover of the preceding vehicle, trajectory predicting, trajectory planning