基于强化学习算法的自适应巡航控制系统测试场景生成方法

打开文本图片集

【欢迎引用】，高艳，秦熠琳，等.基于强化学习算法的自适应巡航控制系统测试场景生成方法[J].汽车文摘，2026（2）： 32-36.

【Cite this paper】ZHAODY，GAO Y，QINYL，etal.AdaptiveCruise Control System Test Scenario Generation Method Basedon ReinforcementLearningAlgorithm[J].AutomotiveDigest （Chinese），2026（2）： 32-36.

关键词：自适应巡航控制系统；测试场景生成；强化学习；DDPG算法中图分类号：U461.91 文献标志码：A DOI：10.19822/j.cnki.1671-6329.20250156

Adaptive Cruise Control System Test Scenario Generation Method Based on ReinforcementLearning Algorithm

ZhaoDeyin’，Gao Yan1，QinYilin'，ZhangPeixing²，ZhangJiaming1 （1.Global R&D Center， China FAW Corporation Limited， Changchun 130o13; 2.Colege of Automotive Engineering， Jilin University， Changchun 130025）

【Abstract】In order to improve the testing efficiencyand comprehensiveness of the Adaptive Cruise Control （ACC） system，thispaper proposes a test scenario generation methodbased on the Deep Deterministic PolicyGradient（DDPG） algorithm.Firstly，typialdrivingsenariosaredesigned，andatestsenariomodelisonstructed，whichincludeseari initialization，thedefinitionof stateandcontrolvariables，theagent-environmentinteractionmechanism，andareward function.Then，anagentisdesignedbasedonareinforcementlearning algorithm，andgeneratechallnging testscenarios.a rewardfunction integrating exposurerate，adversarial factors，andunreasonablecollsion penaltiesisdeveloped，guiding the agenttogeneratescenariosthatarenotonlyhighlyriskybutalsoconsistentwithrealtrafficrules.Finally，simulation experimentsonleft/rightcut-inandcut-outscenariosareconductedtovalidate theefectivenessof theproposed method. Theresultsdemonstrate thatthisapproachcanautomaticallygenerate critical hazardousscenariosforACC，improve test coverage and specificity，and provide significant value for ensuring system safety testing.

Key Words：Adaptive cruise control system，Test scenario generation，Reinforcement learning， DDPG algorithm

0引言

自适应巡航控制系统（AdaptiveCruiseControlSystem，ACC）是自动驾驶辅助系统中的关键功能，通过在纵向自动调节车速和横向保持车辆在车道内行驶，实现车辆智能跟驰和车道保持等L2级自动驾驶能力。（剩余6137字）