Recovery Motion Learning for Arm Mounted Mobile Crawler Robot in Drive System's Failure*. Issue 1 (July 2017)

• Record Type: Journal Article
• Authors: Ito, Tasuku; Kono, Hitoshi; Tamura, Yusuke; Yamashita, Atsushi; Asama, Hajime
• Is Part Of: IFAC-PapersOnLine. Volume 50:Issue 1(2017)
• Page Start: 2329
• Page End: 2334
• Publication Date: 2017-07
• Subjects: Reinforcement learning -- Fault tolerance -- Crawler robot -- Teleoperation -- Motion control -- Intelligent driver aids -- Normalized energy stability margin; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2017.08.475
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: In the disaster area, an arm mounted crawler robot is leveraged for missions such as searching victims. However, the robot system has possibility of failure of drive system at the extreme environment. Moreover, the robot needs to keep moving to repair the mechanism, if the drive system becomes failure. In response to this problem, it is important to realize the recovery motion. However, designing of recovery motion is difficult because the recovery motion depends on the environments and configurations of the robot. This paper describes the learning methodology of the recovery motion in the single-arm mounted crawler robot, and we confirmed that the proposed system can learn the recovery motion in computer simulation.