MoRE: Mixture of Residual Experts for Humanoid Lifelike Gaits Learning on Complex Terrains
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916791691771904 |
|---|---|
| author | Wang, Dewei Wang, Xinmiao Liu, Xinzhe Shi, Jiyuan Zhao, Yingnan Bai, Chenjia Li, Xuelong |
| author_facet | Wang, Dewei Wang, Xinmiao Liu, Xinzhe Shi, Jiyuan Zhao, Yingnan Bai, Chenjia Li, Xuelong |
| contents | Humanoid robots have demonstrated robust locomotion capabilities using Reinforcement Learning (RL)-based approaches. Further, to obtain human-like behaviors, existing methods integrate human motion-tracking or motion prior in the RL framework. However, these methods are limited in flat terrains with proprioception only, restricting their abilities to traverse challenging terrains with human-like gaits. In this work, we propose a novel framework using a mixture of latent residual experts with multi-discriminators to train an RL policy, which is capable of traversing complex terrains in controllable lifelike gaits with exteroception. Our two-stage training pipeline first teaches the policy to traverse complex terrains using a depth camera, and then enables gait-commanded switching between human-like gait patterns. We also design gait rewards to adjust human-like behaviors like robot base height. Simulation and real-world experiments demonstrate that our framework exhibits exceptional performance in traversing complex terrains, and achieves seamless transitions between multiple human-like gait patterns. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_08840 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | MoRE: Mixture of Residual Experts for Humanoid Lifelike Gaits Learning on Complex Terrains Wang, Dewei Wang, Xinmiao Liu, Xinzhe Shi, Jiyuan Zhao, Yingnan Bai, Chenjia Li, Xuelong Robotics Humanoid robots have demonstrated robust locomotion capabilities using Reinforcement Learning (RL)-based approaches. Further, to obtain human-like behaviors, existing methods integrate human motion-tracking or motion prior in the RL framework. However, these methods are limited in flat terrains with proprioception only, restricting their abilities to traverse challenging terrains with human-like gaits. In this work, we propose a novel framework using a mixture of latent residual experts with multi-discriminators to train an RL policy, which is capable of traversing complex terrains in controllable lifelike gaits with exteroception. Our two-stage training pipeline first teaches the policy to traverse complex terrains using a depth camera, and then enables gait-commanded switching between human-like gait patterns. We also design gait rewards to adjust human-like behaviors like robot base height. Simulation and real-world experiments demonstrate that our framework exhibits exceptional performance in traversing complex terrains, and achieves seamless transitions between multiple human-like gait patterns. |
| title | MoRE: Mixture of Residual Experts for Humanoid Lifelike Gaits Learning on Complex Terrains |
| topic | Robotics |
| url | https://arxiv.org/abs/2506.08840 |