复杂交通场景智能驾驶决策研究综述

中图分类号：U463.6 文献标识码：A 文章编号：1003-8639（2025）11-0073-03

Reviewof IntelligentDrivingDecision-makingResearch in Complex Traffic Scenario:

Zhong Chenzhipeng

(School ofAutomotive and Aviation,Jiangxi Technical Collge of Manufacturing,Nanchang 33oo95,China)

【Abstract】To sort out the research contextof intellgent driving decision-making in complex trafic scenarios,this paper conducts asystematic inductionand analysis from thre core dimensions:scenario clasification,decision-making algorithms,and decision-making indicators based on existing research results.According tothediferences in influencing factors,complex traficscenariosaredividedintothreecategories:complexvehiclebehaviorscenarios,hig-densitytraffic flow scenariosandcomplex structuralroad scenarios,andtherisk characteristicsofeach typeof scenarioareanalyzed. Combined with scenariocharacteristics,decision-makingalgorithmsareclasified intolearning-based,interactive,and end-to-end types,and the advantagesand limitations of diferent algorithms inpractical applications are comparedand analyzed.Decision-making indicators are sortedoutfromthree aspects:safety,comfort,and eficiency,andthecore focus indicators under diferent scenariosare clarified.Through theabove analysis,this paper summarizes thekey directions of current research,andatthe same time pointsoutthree majorresearch dificulties,namely insufficient researchon decision-making incompound complex scenarios,lack ofself-evolutionary decision-making mechanisms,and inconsistent decision-making evaluation indicators.Based on this,the paper proposes thre future research directions,including the constructionofmulti-objectivedecision-making models,thedevelopmentofself-learningdecision-making algorithms,and researchoncognition-baseddecision-making,iming toprovidetheoreticalsupportandpracticalreferencesforinteligent driving decision-making research in complex traffic scenarios and future dilemma scenarios.

【Key words】 complex traffc scenarios; intelligent driving decision-making; decision-making algorithms; decision-making evaluation indicators

0 引言

当前智能驾驶系统通过环境感知、运动控制等实现辅助驾驶功能，可以应对常规交通场景，但面对复杂交通场景则暴露出不足，特别是在场景语义理解和复杂场景决策方面。（剩余5113字）