Imitation Learning for High Precision Peg-in-Hole Tasks

Gubbi, S and Kolathaya, S and Amrutur, B (2020) Imitation Learning for High Precision Peg-in-Hole Tasks. In: 6th International Conference on Control, Automation and Robotics ICCAR 2020, 20-23, April 2020, Singapore, pp. 368-372.

PDF int_con_con_aut_rob_368-372_2020.pdf - Published Version Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Industrial robot manipulators are not able to match the precision and speed with which humans are able to execute contact rich tasks even to this day. Therefore, as a means to overcome this gap, we demonstrate generative methods for imitating a peg-in-hole insertion task in a 6-DOF robot manipulator. In particular, generative adversarial imitation learning (GAIL) is used to successfully achieve this task with a 6 μ\mathrmm peg-hole clearance on the Yaskawa GP8 industrial robot. Experimental results show that the policy successfully learns within 20 episodes from a handful of human expert demonstrations on the robot (i.e., < 10 tele-operated robot demonstrations). insertion time improves from > 20 seconds (which also includes failed insertions) to < 15 seconds, thereby validating the effectiveness of this approach.

Item Type:	Conference Paper
Publication:	2020 6th International Conference on Control, Automation and Robotics, ICCAR 2020
Publisher:	Institute of Electrical and Electronics Engineers Inc.
Additional Information:	The copyright of this article belongs to Institute of Electrical and Electronics Engineers Inc.
Keywords:	Agricultural robots; Flexible manipulators; Industrial manipulators; Modular robots; Robot applications; Robotics, 6-dof robots; Generative methods; High-precision; Human expert; Imitation learning; Peg-in-hole tasks; Robot manipulator; Teleoperated robots, Industrial robots
Department/Centre:	Division of Interdisciplinary Sciences > Robert Bosch Centre for Cyber Physical Systems
Date Deposited:	24 Aug 2020 09:27
Last Modified:	24 Aug 2020 09:27
URI:	http://eprints.iisc.ac.in/id/eprint/65989

Actions (login required)

View Item