计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (7): 29-31.DOI: 10.3778/j.issn.1002-8331.2009.07.009

• 博士论坛 • 上一篇    下一篇

局部相对位姿空间内轮式月球车的避障策略学习

潘海宁1,崔平远1,2,居鹤华2   

  1. 1.北京工业大学 电子信息与控制工程学院,北京 100124
    2.哈尔滨工业大学 深空探测基础研究中心,哈尔滨 150001
  • 收稿日期:2008-11-17 修回日期:2008-12-26 出版日期:2009-03-01 发布日期:2009-03-01
  • 通讯作者: 潘海宁

Wheeled lunar rover’s obstacle avoidance strategy learning in local relative pose space

PAN Hai-ning1,CUI Ping-yuan1,2,JV He-hua2   

  1. 1.Department of Electronic Information and Control Engineering,Beijing University of Technology,Beijing 100124,China
    2.Deep Space Exploration Research Center,Harbin Institute of Technology,Harbin 150001,China
  • Received:2008-11-17 Revised:2008-12-26 Online:2009-03-01 Published:2009-03-01
  • Contact: PAN Hai-ning

摘要: 提出了一种轮式月球车的避障学习方法。首先列出了BH2月球车的动力学方程,并将避障行为解释为沿子目标点行走的过程。然后在月球车视觉局部范围内进行避障策略学习,选择车体、子目标点和障碍之间的相对位置矢量为学习过程的状态量,使策略学习对环境变化有鲁棒性;选择车轮的转向力矩为控制输入,降低了学习复杂度。实验证明此方法对环境变化有很好的适应能力。

关键词: 月球车, 避障, 相对位姿, 机器学习

Abstract: A learning approach for the lunar rover’s obstacle avoidance is presented.Firstly the dynamics of the rover BH2 is modeled.The obstacle is avoided by following the sub-goals sequence in the vision range.The relative position vectors among the rover’s body,the sub-goals and the obstacle are chosen to compose the learning states in order to ensure the strategy’s robustness.The rotational torque,which simplifies the learning process,is chosen as the learning action.Experiment results show that the learnt strategy is adaptive to the changing environment.

Key words: lunar rover, obstacle avoidance, relative pose, machine learning