多约束下车辆自主决策可达集分析  

Analysis of Reachable Sets for Autonomous Vehicle Decision-making Under Multiple Constraints

作　　者：杨旭 杨海洋 Yang Xu;Yang Haiyang(Department of Transportation,Xi’an Institute of Transportation Engineering,Xi’an 710065,China)

机构地区：[1]西安交通工程学院交通运输系,西安710065

出　　处：《黑龙江科学》2023年第24期89-91,共3页Heilongjiang Science

基　　金：陕西省教育厅科学研究计划项目资助(22JK0452)。

摘　　要：探究了自主车辆安全决策中存在的问题,包括复杂的操作条件及不确定性,传统验证方法无法满足这些需求,会降低自主车辆的安全性。为解决这一问题,提出了一种创新方法,利用车道边界作为主要约束条件,运用哈密顿-雅可比方程管理可达集,通过水平集技术可视化可达集,帮助自主车辆界定安全区域。研究结果表明,此方法提高了自主车辆的适应性,为其在复杂道路环境中的智能驾驶提供了理论支持,提升了道路安全性。这一多目标约束可达集建模方法有望推动自主驾驶汽车技术的进一步发展,实现更安全高效的道路交通。The study discusses the challenges in autonomous vehicle safety decision-making,including complex operating conditions and uncertainties.Traditional validation methods can not satisfy these demands,and have reduced confidence in the safety decisions of autonomous vehicles.To address this issue,the study proposes an innovative approach that uses lane boundaries as the primary constraint,employs Hamilton-Jacobi equations to manage reachable sets,and visualizes reachable sets through level set techniques to help autonomous vehicles define safe regions.Research results indicate that this approach enhances the adaptability of autonomous vehicles,and provides theoretical support for intelligent driving in complex road environments,thereby improving safety on the roads.This multi-objective constraint reachability set modeling approach is expected to drive further advancements in autonomous vehicle technology,leading to safer and more efficient future road traffic.

关 键 词：智能交通 安全性验证 可达集建模 安全决策 

分 类 号：X913.3[环境科学与工程—安全科学] U471.15[机械工程—车辆工程]