欢迎访问《兵工学报》官方网站,今天是 分享到:

兵工学报 ›› 2017, Vol. 38 ›› Issue (12): 2301-2308.doi: 10.3969/j.issn.1000-1093.2017.12.002

• 论文 • 上一篇    下一篇

基于高斯混合-隐马尔可夫模型的速差转向履带车辆横向控制驾驶员模型

王博洋, 龚建伟, 高天云, 陈慧岩, 席军强   

  1. (北京理工大学 机械与车辆学院, 北京 100081)
  • 收稿日期:2017-07-20 修回日期:2017-07-20 上线日期:2018-02-01
  • 通讯作者: 陈慧岩(1961—),男,教授,博士生导师 E-mail:chen_h_y@263.net
  • 作者简介:王博洋(1991—),男,博士研究生。E-mail:wbythink@163.com
  • 基金资助:
    武器装备预先研究项目(ZLY2015315)

Steering Control Driver Model of Skid Steering Vehicle Based on Gaussian Mixture Model-Hidden Markov Model

WANG Bo-yang, GONG Jian-wei, GAO Tian-yun, CHEN Hui-yan, XI Jun-qiang   

  1. (School of Mechanical Engineering, Beijing Institute of Technology, Beijing 100081, China)
  • Received:2017-07-20 Revised:2017-07-20 Online:2018-02-01

摘要: 为解决基于离合器转向机的履带车辆在无人行驶条件下的横向控制问题,采用一种基于高斯混合-隐马尔可夫模型的统计学习方法构建驾驶员模型,以实现对驾驶员跟踪控制操控经验的表述。利用经过大量试验采集获得的经验驾驶员操控数据对模型进行训练。以基于高斯混合模型表征的车辆速度和航向偏差作为隐马尔可夫模型的观测状态参量,并利用高斯混合模型对左右操纵杆位置进行转向模式划分,以转向模式作为隐马尔可夫模型的隐藏层状态参量,通过对模型的训练最终实现对于驾驶员操控经验以及车辆特性的统计学描述。利用上述模型对跟踪控制过程中的期望转向模式进行预测分析,结果表明该模型能够较准确地对转向模式进行预测。

关键词: 兵器科学与技术, 履带车辆, 横向控制, 驾驶员模型, 高斯混合-隐马尔可夫模型, 机器学习, 运动基元

Abstract: In order to solve the unmanned lateral control problem of the skid-steering vehicle based on clutch and brake steering structure, the GMM-HMM model is used to predict the steering mode. The skilled driver's steering operation data acquired from the numerous filed tests is applied to establish the model. The observation states of the HMM model are made up of the velocity and the heading deviation based on the GMM model. The hidden states of the HMM model are made up of the cluster labels of the steering stick position including both of the left and the right sides. The driver-vehicle interaction model of the skid-steering vehicle based on clutch and brake steering structure is established from data training. The driving skills and the vehicle dynamics are described in the statistics way. The model is applied to estimate the steering mode, and the results have proved that the steering mode can be estimated properly based on the driving skills. Key

Key words: ordnancescienceandtechnology, skidsteering, steeringcontrol, drivermodel, Gaussianmixturemodel-hiddenMarkovmodel, machinelearning, motionprimitive

中图分类号: