基于反馈补偿的汇入场景下车辆模型预测控制

打开文本图片集

主题词：模型预测控制反馈补偿智能车辆汇入场景鲁棒性

中图分类号：U462.1 文献标志码：A DOI：10.19620/j.cnki.1000-3703.20240979

Model Predictive Control of Vehicles in Merging Scenarios Based on Feedback Compensation

Du Zhibin1²，Niu Haoran³，Wang Jiangbo²，Yuan Shoutong1²，XieHuil （1.Schoolof MechanicalEngineering，Tianjin University，Tianjin3Oo350;2.CATARC IntellgentandConnected Technology CompanyLimited，Tianjin3Oo3oo;3.SchoolofInfrastructureEngieering，Dalian UniversityofTechology， Dalian ）

【AbstractInresponsetothehighrobustness and stabilityrequirements forintellgent vehicle trajectory tracking control inexpresswaymergingareas，this studydesignsavehicle trajectorytrackingcontrolalgorithmbasedonthefusionof KMPCPIDmulti-controllers.ThisfusedpredictivecontrolerisbuiltuponModelPredictControl（MPC）control，withProportional Integral Derivative（PI）feedback control serving asthecontrolcompensation，andthecontroller parametersare tuned based onreal merging scenarios.Both CARLA software simulationand Hardware-In-the-Loop（HIL）simulation are employed to verifytheeffectivenessandrobustnessof thecontroler.ThesimulationresultsdemonstratethattheKMPC-PIDcontroller performs wellunder various testconditions in urban expressway merging scenarios.Compares to the MPC without feedback control，inCARLAsimulations，themaximumlateralabsolutedviationdecreasesby29.75%，theaveragelateraldeviation decreasesby7.14%，themaximumlongitudinalabsolutedeviationdecreasedby64.24%，andtheaveragelongitudinaldeviation decreasesby 65.56% .InHILsimulations，although theaverage diferences inlateral and longitudinal deviationsare relatively small，the proposed controler stillachievesa21.33%reduction in themaximum lateralabsolutedeviationanda1.63 % （204 reduction in the maximumlongitudinal absolute deviation.

Keywords：Model Predict Control （MPC），Feedback compensation，Inteligent vehicle， Merging scenarios，Robustness

【引用格式】杜志彬，牛浩然，王江波，等.基于反馈补偿的汇入场景下车辆模型预测控制[J].汽车技术，2025（10）：32-40.DU ZB，NIUHR，WANGJB，etal.ModelPredictive ControlofVehiclesin Merging ScenariosBasedonFeedbackCompensation[J]. Automobile Technology，2025（10）： 32-40.

1前言

在城市快速路合流区汇入场景中，车辆冲突集中、速度差异大，外部干扰更强，安全风险高[1-2]，因而控制车辆精准完成轨迹跟踪是实现安全汇入的关键。（剩余13440字）