基于广义动态模糊神经网络的自动驾驶换道策略优化方法  

Generalized Dynamic Fuzzy Neural Network Enabled Lane Changing Strategy Optimization Method for Autonomous Vehicles

作　　者：王嘉文 胡晨曦 李少波 WANG Jia-wen;HU Chen-xi;LI Shao-bo(Business School,University of Shanghai for Science and Technology,Shanghai 200093,China;Shanghai Minhang Urban Infrastructure Construction Investment Development Co.,Ltd,Shanghai 201199,China)

机构地区：[1]上海理工大学管理学院,上海200093 [2]上海闵行城市建设投资开发有限公司,上海201199

出　　处：《系统工程》2022年第6期113-120,共8页Systems Engineering

基　　金：国家自然科学基金青年科学基金资助项目(52102398);上海市软科学研究项目(22692194300)。

摘　　要：为使自动驾驶车辆行驶效率提高,研究安全且高效的自动驾驶车辆换道策略是基础问题之一。本文基于广义动态模糊神经网络模型提出一种以运行效率为优化目标的自动驾驶换道策略。首先,基于元胞自动机建立高速公路交通流模型,分别实现人工驾驶、自动驾驶车辆的跟驰与换道表达方法;进而设计并实现了基于广义动态模糊神经网络(GD-FNN)的自学习算法优化自动驾驶车辆换道策略;最后与元胞自动机经典换道策略进行对比。结果表明:GD-FNN换道策略在处于低密度区具有更高的换道频率;GD-FNN换道策略在驾驶效率上整体优于STCA,在中低密度条件下单位时间内行驶距离可提升12.88%.In order to improve the driving efficiency of autonomous vehicles, one of the basic problems is to realize the lane changing strategy safely and efficiently. Based on the generalized dynamic fuzzy neural network(GD-FNN) model, this paper proposes an automatic driving lane changing strategy with the optimization goal of operation efficiency. Firstly, the traffic flow model of expressway is established based on cellular automata, and the car following and lane changing models for manual driving and automatic driving vehicles are realized respectively. Secondly, a self-learning algorithm based on generalized dynamic fuzzy neural network is designed and implemented to optimize the lane changing strategy of autonomous vehicles. Finally, proposed lane changing strategy is compared with the classical lane changing strategy of cellular automata. The results show that the GD-FNN lane change strategy has higher lane change frequency in the low-density area. The overall driving efficiency of GD-FNN lane changing strategy is better than STCA, and the driving distance per unit time can be increased by 12.88% under medium and low traffic density conditions.

关 键 词：自动驾驶汽车 换道策略 高效驾驶 广义动态模糊神经网络 

分 类 号：U491[交通运输工程—交通运输规划与管理]