混行场景智能车换道决策与运动规划

Intelligent vehicle lane change decision and motion planning in mixed traffic scenarios

针对行人-车辆混行的常见交通场景下智能车决策安全性和行驶效率不高的问题,提出了一种新的基于自车期望车速与前车车速、加速度和车距的行车不满意度换道行为决策模型。同时建立换道最小安全距离模型,用以在换道全过程中判断换道的可行性。为了提高运动规划算法的效率,选用Frenet坐标系,采用路径规划和速度规划解耦的方式。对于路径规划,选择五次多项式曲线,采用考虑安全、舒适以及高效性的3个路径评估指标。对于速度规划,采用动态规划与二次规划获得平滑的速度曲线。在CarSim/PreScan/Simulink的联合仿真平台下搭建人车混行的交通场景进行验证。仿真结果表明,基于行车不满意度的换道决策模型能选择更高效及安全的行驶方式,运动规划模块能确保自车换道及避让行人过程的安全性和操纵稳定性。

To address the safety risks and low driving efficiency in mixed traffic scenarios with both pedestrians and vehicles,this paper proposes a novel lane-changing decision-making model for intelligent vehicles.Based on the dissatisfaction of the ego vehicle,it considers its desired speed,the speed of the preceding vehicle,acceleration,and following distance.Meanwhile,a minimal safety distance model for lane changing is built to assess the feasibility of lane changing throughout the entire process.To enhance the real-time performance of motion planning algorithms,the Frenet coordinate system is chosen,and a decoupled approach for path planning and velocity planning is employed.For path planning,a fifth-order polynomial curve is selected,incorporating three evaluation criteria:safety,comfort,and efficiency.Dynamic programming combined with quadratic programming is utilized for velocity planning to obtain a smooth speed profile.Finally,a joint simulation platform using CarSim/PreScan/Simulink is employed to validate the proposed model in a mixed traffic scenario.Our simulation results show the lane-changing decision-making model based on driving dissatisfaction effectively selects more efficient and safer driving strategies.The motion planning module ensures the safety and stability of the ego vehicle when changing lanes or shunning pedestrians.