基于用户画像及多车协同的环岛车辆换道策略
Vehicle Lane Change Strategy in Roundabout Based on Vehicle User Profile and Multi-Vehicle Collaboration
曹栋发 1胡创业 2丁男1
作者信息
- 1. 新疆师范大学计算机科学技术学院,新疆 乌鲁木齐 830054;大连理工大学工业装备智能控制与优化教育部重点实验室,辽宁 大连 116024
- 2. 新疆师范大学计算机科学技术学院,新疆 乌鲁木齐 830054
- 折叠
摘要
基于智能网联汽车车-路-环境的协同策略已成为一种解决智能交通领域问题的有效方案.因此通过结合交通环岛的行车特点,提出了一种基于车辆用户画像的交通环岛换道策略.首先为了解决交通环岛中智能网联汽车多源异构数据的混杂问题,定义了车辆用户画像,对智能网联汽车多维数据进行表征.其次,基于随机森林算法对车辆用户画像中标签的权重进行实时更新.再次,根据车辆用户画像及动态权重设计收益函数针对环岛出口位置的内道车辆进行换道决策,实现了决策的场景自适应.最后,利用SUMO对本算法在交通拥堵、正常、稀疏等三种场景中进行仿真测试.实验结果表明本算法使得车辆在交通环岛的通行效率和舒适度均得到有效提升,以此验证了车辆用户画像对于优化决策的场景自适应效果的作用.
Abstract
The collaborative strategy of vehicle-road-environment based on intelligent and connected vehicles(ICVs)has become an effective solution to solve the problems in intelligent transportation.This paper proposes a strategy of vehicle lane change based on Vehicle User Profile in combination with the driving characteristics of round-abouts.Firstly,in order to solve the confusion problem of multi-source heterogeneous data of ICVs in roundabouts,the Vehicle User Profile was defined to describe and characterize the multidimensional data of ICVs.Secondly,the weights of relevant parameters in the vehicle user profile were updated based on the random forest algorithm.Thirdly,according to the vehicle user profile and dynamic weight,the payoff function was designed to make lane change deci-sions for the inner lane vehicles at the exit position of the roundabout,then realized the scene adaptation of decision-making.Finally,three scenarios were used in the simulation verification,such as traffic congestion,normal and sparse.The experimental results show that this algorithm can effectively improve the efficiency of vehicle traffic in round-abouts.In particular,the efficiency and comfort of vehicles in roundabouts are effectively improved in normal traffic scenarios.The effect of the vehicle user profile on the scene adaptive effect of optimal decision-making is verified.
关键词
智能交通/车辆用户画像/随机森林算法/环岛/动态权重Key words
Intelligent transportation/Vehicle user profile/Random forest algorithm/Roundabouts/Dynamic weight引用本文复制引用
基金项目
新疆维吾尔自治区自然科学基金杰出青年科学基金(2021D01E20)
国家自然科学基金(62072071)
出版年
2024
NETL
NSTL
万方数据