Robot reinforcement learning on the constraint manifold
Liu, Puze ; Tateo, Davide LU
; Bou-Ammar, Haitham
and Peters, Jan
(2021)
5th Conference on Robot Learning, CoRL 2021
In Proceedings of Machine Learning Research
164.
p.1357-1366
- Abstract
Reinforcement learning in robotics is extremely challenging due to many practical issues, including safety, mechanical constraints, and wear and tear. Typically, these issues are not considered in the machine learning literature. One crucial problem in applying reinforcement learning in the real world is Safe Exploration, which requires physical and safety constraints satisfaction throughout the learning process. To explore in such a safety-critical environment, leveraging known information such as robot models and constraints is beneficial to provide more robust safety guarantees. Exploiting this knowledge, we propose a novel method to learn robotics tasks in simulation efficiently while satisfying the constraints during the learning... (More)
Reinforcement learning in robotics is extremely challenging due to many practical issues, including safety, mechanical constraints, and wear and tear. Typically, these issues are not considered in the machine learning literature. One crucial problem in applying reinforcement learning in the real world is Safe Exploration, which requires physical and safety constraints satisfaction throughout the learning process. To explore in such a safety-critical environment, leveraging known information such as robot models and constraints is beneficial to provide more robust safety guarantees. Exploiting this knowledge, we propose a novel method to learn robotics tasks in simulation efficiently while satisfying the constraints during the learning process.
(Less)
- author
- Liu, Puze
; Tateo, Davide
LU
; Bou-Ammar, Haitham
and Peters, Jan
- publishing date
- 2021
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- Constrained reinforcement learning, Robot learning, Safe exploration
- host publication
- 5th Conference on Robot Learning, CoRL 2021
- series title
- Proceedings of Machine Learning Research
- volume
- 164
- pages
- 10 pages
- conference name
- 5th Conference on Robot Learning, CoRL 2021
- conference location
- London, United Kingdom
- conference dates
- 2021-11-08 - 2021-11-11
- external identifiers
-
- scopus:85171771550
- ISSN
- 2640-3498
- language
- English
- LU publication?
- no
- additional info
- Publisher Copyright: © 2021 Proceedings of Machine Learning Research. All rights reserved.
- id
- fb00d0b5-0762-4b8e-a932-10506ca76cc8
- date added to LUP
- 2025-10-16 14:37:56
- date last changed
- 2025-10-23 03:43:29
@inproceedings{fb00d0b5-0762-4b8e-a932-10506ca76cc8,
abstract = {{<p>Reinforcement learning in robotics is extremely challenging due to many practical issues, including safety, mechanical constraints, and wear and tear. Typically, these issues are not considered in the machine learning literature. One crucial problem in applying reinforcement learning in the real world is Safe Exploration, which requires physical and safety constraints satisfaction throughout the learning process. To explore in such a safety-critical environment, leveraging known information such as robot models and constraints is beneficial to provide more robust safety guarantees. Exploiting this knowledge, we propose a novel method to learn robotics tasks in simulation efficiently while satisfying the constraints during the learning process.</p>}},
author = {{Liu, Puze and Tateo, Davide and Bou-Ammar, Haitham and Peters, Jan}},
booktitle = {{5th Conference on Robot Learning, CoRL 2021}},
issn = {{2640-3498}},
keywords = {{Constrained reinforcement learning; Robot learning; Safe exploration}},
language = {{eng}},
pages = {{1357--1366}},
series = {{Proceedings of Machine Learning Research}},
title = {{Robot reinforcement learning on the constraint manifold}},
volume = {{164}},
year = {{2021}},
}