考虑周车主观侵略性的智能车辆换道优化策略

Optimal lane-changing strategy considering subjective aggressiveness of surrounding vehicles for intelligent vehicles

智能车辆换道策略是智能驾驶技术的重要组成部分,在不同场景下的适应能力对于高品质自动驾驶尤为重要。在异质交通流中,智能车辆认知属性的不同会引起换道车辆换道意图及轨迹规划的差异。该策略利用车辆行驶空间、行驶效率参数建立车辆的决策代价函数,采用模糊理论从速度系数、空间系数、车道分布等角度实现对周围车辆主观侵略性的量化描述,在此基础上,通过智能驾驶员模型(IDM)预测周车对换道车辆换道意图的行为响应,并根据预测信息优化换道决策。在轨迹规划过程中,将车辆换道轨迹解耦为车辆换道路径及行驶速度,利用车辆主观侵略性及车辆行驶安全场评估车辆换道行驶风险,并将行驶安全场与二次规划相融合,从安全性、舒适性、参考线偏差等角度进行路径及速度优化。仿真结果表明,该优化策略通过换道决策可以有效提高道路通行效率,相较固定五次多项式与单独的安全场换道轨迹,可以有效降低换道过程中横向速度与横向加速度的幅值,改善换道过程中的行驶舒适性。

With the increasing number of vehicles in China,driving safety is becoming more and more critical and challenging.Intelligent vehicle lane-changing strategies,encompassing lane-changing behavior decision and trajectory planning,are important to improve the vehicle driving safety.Safe and effective lane change decision and trajectory planning not only guarantees the safety of drivers,but also effectively improves the road traffic efficiency.However,the vehicle lane change decision and trajectory planning demonstrate strong coupling relationship in the study of lane change behaviors.The collaborative design of lane changing decision and trajectory planning for autonomous driving is thus of paramount significance.The adaptability of the vehicle lane-changing strategy to different scenarios is crucial for high-quality autonomous driving.The decision-making module provides the target lane information for the planning module,and the planning module plans the specific driving trajectory according to the decision-making information and road conditions.In heterogeneous traffic flows,variations in the cognitive attribute of intelligent vehicles lead to differences in their lane-changing intentions and trajectory planning.During the vehicle driving process,the social attribute of the driver causes different social interaction behaviors.Therefore,it is particularly important to analyze the behavioral interaction of intelligent vehicles from the perspective of the drivers’social attributes.The social attributes of vehicles mainly include two parts:individualized attributes and social cognitive attributes.The selfishness or altruism of the individualized attribute is mainly used to predict the behavioral interactions of vehicles in heterogeneous traffic flows,while the social cognitive attributes of vehicles are mainly reflected by the tendency to bully the weak and fear the strong.Consequently,the modeling of vehicle risk perception from the driver’s perspective can effectively predict and avoid driving risks in the traffic.Thus,this paper proposes a novel optimal lane-changing strategy with full consideration of subjective aggressiveness of surrounding vehicles for intelligent vehicles.The social cognitive attributes of vehicles utilize the method that combines the traffic safety field with subjective aggression to conduct behavioral decision-making and trajectory planning.Besides,the impacts of surrounding vehicles’operation differences on lane-changing control are investigated.The strategy establishes the vehicle’s decision-making cost function based on the vehicle driving space and efficiency,and utilizes the fuzzy theory to realize the quantitative description of the subjective aggressiveness of the surrounding vehicles in terms of vehicle speed coefficients,spatial coefficients,and lane distributions.On this basis,the intelligent driver model(IDM)is adopted to predict the behavioral response of the surrounding vehicles to the lane-changing intention of the ego vehicle.The lane-changing decision-making strategy is further optimized based on the predicted information.In the trajectory planning,the vehicle lane-changing trajectory is decoupled into the vehicle lane-changing path and driving speed.Discrete sampling is carried out based on the established driving safety field,and the optimal path planning points are selected by considering the vehicle driving safety and vehicle reference line deviation.The waypoint information is fitted into a polynomial curve for quadratic programming,in which the optimized and smoothed lane change path curve is calculated.In addition to the vehicle lane change path information,it is also necessary to carry out the corresponding speed planning according to the state information of surrounding vehicles.The vehicle’s subjective aggressiveness and driving safety field are utilized to assess lane-changing risks.The driving safety field is integrated with quadratic planning to optimize the path and speed from the perspective of safety,comfort,and deviation from the reference line.Our simulation results show the proposed optimization strategy significantly enhances road throughput via strategic lane-changing decisions.Compared with the fixed fifth-degree polynomial and separate safety field lane-changing trajectories,the proposed strategy markedly reduces the amplitude of lateral velocity and lateral acceleration during lane-changing and improves driving comfort.