Exciting action : investigating efficient exploration for learning musculoskeletal humanoid locomotion
Geiß, Henri Jacques ; Al-Hafez, Firas ; Seyfarth, Andre ; Peters, Jan and Tateo, Davide LU
(2024)
23rd IEEE-RAS International Conference on Humanoid Robots, Humanoids 2024
In IEEE-RAS International Conference on Humanoid Robots
p.205-212
- Abstract
Learning a locomotion controller for a musculoskeletal system is challenging due to over-actuation and high-dimensional action space. While many reinforcement learning methods attempt to address this issue, they often struggle to learn human-like gaits because of the complexity involved in engineering an effective reward function. In this paper, we demonstrate that adversarial imitation learning can address this issue by analyzing key problems and providing solutions using both current literature and novel techniques. We validate our methodology by learning walking and running gaits on a simulated humanoid model with 16 degrees of freedom and 92 Muscle-Tendon Units, achieving natural-looking gaits with only a few demonstrations. Code is... (More)
Learning a locomotion controller for a musculoskeletal system is challenging due to over-actuation and high-dimensional action space. While many reinforcement learning methods attempt to address this issue, they often struggle to learn human-like gaits because of the complexity involved in engineering an effective reward function. In this paper, we demonstrate that adversarial imitation learning can address this issue by analyzing key problems and providing solutions using both current literature and novel techniques. We validate our methodology by learning walking and running gaits on a simulated humanoid model with 16 degrees of freedom and 92 Muscle-Tendon Units, achieving natural-looking gaits with only a few demonstrations. Code is available at https://github.com/henriTUD/musculoco-learning.
(Less)
- author
- Geiß, Henri Jacques
; Al-Hafez, Firas
; Seyfarth, Andre
; Peters, Jan
and Tateo, Davide
LU
- publishing date
- 2024
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- host publication
- 2024 IEEE-RAS 23rd International Conference on Humanoid Robots, Humanoids 2024
- series title
- IEEE-RAS International Conference on Humanoid Robots
- pages
- 8 pages
- publisher
- IEEE Computer Society
- conference name
- 23rd IEEE-RAS International Conference on Humanoid Robots, Humanoids 2024
- conference location
- Nancy, France
- conference dates
- 2024-11-22 - 2024-11-24
- external identifiers
-
- scopus:85214497202
- ISSN
- 2164-0580
- 2164-0572
- ISBN
- 9798350373578
- DOI
- 10.1109/Humanoids58906.2024.10769835
- language
- English
- LU publication?
- no
- id
- e15dc91f-8846-4d16-b179-407f5ed8401e
- date added to LUP
- 2025-10-16 14:12:35
- date last changed
- 2025-10-30 15:12:25
@inproceedings{e15dc91f-8846-4d16-b179-407f5ed8401e,
abstract = {{<p>Learning a locomotion controller for a musculoskeletal system is challenging due to over-actuation and high-dimensional action space. While many reinforcement learning methods attempt to address this issue, they often struggle to learn human-like gaits because of the complexity involved in engineering an effective reward function. In this paper, we demonstrate that adversarial imitation learning can address this issue by analyzing key problems and providing solutions using both current literature and novel techniques. We validate our methodology by learning walking and running gaits on a simulated humanoid model with 16 degrees of freedom and 92 Muscle-Tendon Units, achieving natural-looking gaits with only a few demonstrations. Code is available at https://github.com/henriTUD/musculoco-learning.</p>}},
author = {{Geiß, Henri Jacques and Al-Hafez, Firas and Seyfarth, Andre and Peters, Jan and Tateo, Davide}},
booktitle = {{2024 IEEE-RAS 23rd International Conference on Humanoid Robots, Humanoids 2024}},
isbn = {{9798350373578}},
issn = {{2164-0580}},
language = {{eng}},
pages = {{205--212}},
publisher = {{IEEE Computer Society}},
series = {{IEEE-RAS International Conference on Humanoid Robots}},
title = {{Exciting action : investigating efficient exploration for learning musculoskeletal humanoid locomotion}},
url = {{http://dx.doi.org/10.1109/Humanoids58906.2024.10769835}},
doi = {{10.1109/Humanoids58906.2024.10769835}},
year = {{2024}},
}