Hung, C.-M., Sun, L. orcid.org/0000-0002-0393-8665, Wu, Y. et al. (2 more authors) (2021) Introspective visuomotor control : exploiting uncertainty in deep visuomotor control for failure recovery. In: ICRA 2021 : IEEE International Conference on Robotics and Automation. 2021 IEEE International Conference on Robotics and Automation (ICRA), 30 May - 05 Jun 2021, Xi’an, China. Institute of Electrical and Electronics Engineers , pp. 6293-6299. ISBN 978-1-7281-9077-8
Abstract
End-to-end visuomotor control is emerging as a compelling solution for robot manipulation tasks. However, imitation learning-based visuomotor control approaches tend to suffer from a common limitation, lacking the ability to recover from an out-of-distribution state caused by compounding errors. In this paper, instead of using tactile feedback or explicitly detecting the failure through vision, we investigate using the uncertainty of a policy neural network. We propose a novel uncertainty-based approach to detect and recover from failure cases. Our hypothesis is that policy uncertainties can implicitly indicate the potential failures in the visuomotor control task and that robot states with minimum uncertainty are more likely to lead to task success. To recover from high uncertainty cases, the robot monitors its uncertainty along a trajectory and explores possible actions in the state-action space to bring itself to a more certain state. Our experiments verify this hypothesis and show a significant improvement on task success rate: 12% in pushing, 15% in pick-and-reach and 22% in pick-and-place.
Metadata
Item Type: | Proceedings Paper |
---|---|
Authors/Creators: |
|
Copyright, Publisher and Additional Information: | © 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works. Reproduced in accordance with the publisher's self-archiving policy. |
Keywords: | Uncertainty; Monte Carlo methods; Conferences; Neural networks; Tactile sensors; End effectors; Trajectory |
Dates: |
|
Institution: | The University of Sheffield |
Academic Units: | The University of Sheffield > Faculty of Engineering (Sheffield) > Department of Computer Science (Sheffield) |
Funding Information: | Funder Grant number Engineering and Physical Science Research Council EP/R026092/1 The Royal Society RGS\R2\202432 |
Depositing User: | Symplectic Sheffield |
Date Deposited: | 20 May 2021 07:44 |
Last Modified: | 21 Jun 2023 15:24 |
Status: | Published |
Publisher: | Institute of Electrical and Electronics Engineers |
Refereed: | Yes |
Identification Number: | 10.1109/ICRA48506.2021.9561749 |
Related URLs: | |
Open Archives Initiative ID (OAI ID): | oai:eprints.whiterose.ac.uk:174297 |