基于神经网络的无人驾驶车辆转向控制研究  

Research on Unmanned Vehicle Steering Control Based on Neural Network

作　　者：曹艳玲 乔梦楠 Cao Yanling;Qiao Mengnan(College of Vehicle and Traffic Engineering,Henan University of Science and Technology,Luoyang City,Henan Province 471003,China)

机构地区：[1]河南科技大学车辆与交通工程学院

出　　处：《农业装备与车辆工程》2019年第5期80-83,共4页Agricultural Equipment & Vehicle Engineering

摘　　要：为了提高无人驾驶车辆进行路径跟踪时转向的准确性,基于神经网络控制理论,利用ADAMS/Car与MATLAB/Simulink进行无人驾驶车辆转向控制联合仿真。利用ADAMS/Car模块建立整车模型,进行规定路径下的跟踪实验并收集路径、车速、前轮转角等信息,以作为神经网络的训练样本。利用MATLAB对训练样本进行训练,并在Simulink中建立神经网络控制器。最后利用ADAMS/Control模块连接ADAMS/Car与MATLAB/Simulink,实现无人驾驶车辆路径跟踪时转向控制的联合仿真。仿真分析结果表明:所建立的神经网络转向控制器能够对路径进行良好的跟踪且具有良好的鲁棒性;同时验证了联合仿真的可行性与优越性,为智能车辆的整车开发提供了思路。In order to improve the accuracy of steering in path tracking of unmanned vehicle,the co-simulation of steering control of unmanned vehicle is carried out by using ADAMS/Car and MATLAB/Simulink,based on neural network control theory.The vehicle model is built by ADAMS/Car module,and the tracking experiments under the prescribed path are carried out,then,information such as path, speed and front wheel corner is collected as training samples of neural network.The training samples are trained by MATLAB,and the neural network controller is established in Simulink.Finally,the ADAMS/Control module is used to connect ADAMS/Car and MATLAB/ Simulink to realize co-simulation of steering control for unmanned vehicle path tracking.The simulation results show that the established neural network steering controller with good robustness can track the path well.At the same time,the feasibility and superiority of the cosimulation are verified,which provide a thought for the development of intelligent vehicle.

关 键 词：神经网络 无人驾驶车辆 转向控制 ADAMS/CAR MATLAB/SIMULINK 联合仿真 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] U463.4[自动化与计算机技术—控制科学与工程]