DatadrivingReinforcementLearningonthePathPlanningforAutonomousVehicles

DatadrivingReinforcementLearningonthePathPlanningforAutonomousVehicles.pdf
Thepathplanningforautonomousvehiclesisahottopicinacademicworld.Thegoalofthisproblemistodesignavehi-clewithlearningabilitytoapproachthetargetwithoutanycollision.Inthispaper,wefocusondata-drivingreinforcementlearning(RL)designforthepathplanningforautonomousvehiclesproblems.Weproposedthemethodofsensordetection,andaself-learningstrategyforavehicleseekingthetargetwithobstacleavoidance.Specifically,wedesignedacontinuousreinforcementsignaltoimprovethesystem'spreferentialdecisionbetweenthetargetseekingandtheobstacleavoidance.Toverifythelearningabilityofourstrategy,wedevelopedanin-doorenvironmentwithdifferentexperiments.ThesimulationresultsshowthattheRLpresentsaneffectivelearningabilityforthepathplanningforautonomousvehiclesproblems.
作者：FangXiao XiePing GaoHongbo HeQun LiNaiyi
作者单位：CheryScientificResearchInstituteWuhu;ShanghaiJiaoTongUniversityYanshanUniversityCheryScientificResearchInstituteWuhu;NanjingUniversityofScienceandTechnologyCheryScientificResearchInstituteWuhu
母体文献：2016中国汽车工程学会年会论文集
会议名称：2016中国汽车工程学会年会  
会议时间：2016年10月26日
会议地点：上海
主办单位：中国汽车工程学会
语种：chi
分类号：
关键词：autonomousvehicle  pathplanning  reinforcementlearning  adaptivedynamicprogramming
在线出版日期：2019年4月25日
基金项目：
相似文献
相关博文

DatadrivingReinforcementLearningonthePathPlanningforAutonomousVehicles.pdf

文件大小:

2.05 MB

高速下载