摘要
针对智能车辆在复杂动态环境中难以与周围车辆有效交互,换道行为过于保守或激进的问题,提出一种基于博弈的交互决策与轨迹规划算法。根据驾驶员操作特性和驾驶风格建立人类驾驶车辆模型,利用贝叶斯推理算法评估驾驶行为的保守程度,实现对交互车辆博弈动作的实时预测;在此基础上建立自车换道决策规划模型,首先采样换道终点生成换道候选路径,随后构建自车收益函数评价博弈过程中的换道策略,最后设计一种融合主从博弈思想的动态规划算法,考虑车辆间的交互情况下生成最优策略和轨迹。通过Matlab/Simulink和Pre Scan搭建的人在环仿真平台,对多组换道工况进行验证分析,结果表明该算法在复杂动态的换道场景中能够有效与周围车辆进行交互,行为决策合理,运动轨迹在保证安全的基础上兼顾平顺性。
A game-based interactive decision-making and trajectory planning algorithm is proposed to address the problem that overly conservative or aggressive lane-changing behaviors will be exhibited by intelligent vehicles in complex dynamic scenarios due to lake of effective interaction with surrounding vehicles.A human-driven vehicle model is established based on driver operation characteristics and driving styles,by utilizing Bayesian inference algorithms to assess the conservativeness of driving behavior and enabling real-time prediction of interactive vehicle maneuvering actions.On this basis,a decision-making and planning model for lane-changing maneuvers of the ego-vehicle is developed.Initially,lane-changing endpoints are sampled to generate candidate lane-changing paths.Subsequently,an ego-vehicle reward function is constructed to evaluate lane-changing strategies during the game process.Finally,a dynamic programming algorithm is designed,which incorporates the idea of leader-follower games and fully considers the generation of optimal strategies and trajectories in interactive gaming scenarios.The human-in-the-loop simulation platform,built with Matlab/Simulink and PreScan,is utilized to validate and analyze multiple lane-changing scenarios.The results indicate that the algorithm effectively interacts with surrounding vehicles in complex and dynamic lane-changing scenarios,demonstrating rational behavioral decisions and producing trajectories that prioritize both safety and smoothness.
作者
胡杰
赵文龙
郑嘉辰
周思龙
张志凌
吴作伟
陈家骥
HU Jie;ZHAO Wenlong;ZHENG Jiachen;ZHOU Silong;ZHANG Zhiling;WU Zuowei;CHEN Jiaji(Hubei Key Laboratory of Modern Auto Parts Technology,Wuhan 430070;Auto Parts Technology Hubei Collaborative Innovation Center,Wuhan 430070;Hubei Technology Research Center of New Energy and Intelligent Connected Vehicle Engineering,Wuhan 430070)
出处
《机械工程学报》
北大核心
2025年第10期322-334,共13页
Journal of Mechanical Engineering
基金
2023年湖北省重大攻关资助项目(JD)(2023BAA017)。
关键词
智能车辆
换道交互
决策规划
主从博弈
动态规划
intelligent vehicles
lane changing interaction
decision planning
master-slave game
dynamic programming
作者简介
胡杰,男,1984年出生,博士,教授,博士研究生导师。主要研究方向为智能驾驶。E-mail:auto_hj@163.com。
