Meta-Evolve: Continuous Robot Evolution for One-to-many Policy Transfer
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866929336436654080 |
|---|---|
| author | Liu, Xingyu Pathak, Deepak Zhao, Ding |
| author_facet | Liu, Xingyu Pathak, Deepak Zhao, Ding |
| contents | We investigate the problem of transferring an expert policy from a source robot to multiple different robots. To solve this problem, we propose a method named $Meta$-$Evolve$ that uses continuous robot evolution to efficiently transfer the policy to each target robot through a set of tree-structured evolutionary robot sequences. The robot evolution tree allows the robot evolution paths to be shared, so our approach can significantly outperform naive one-to-one policy transfer. We present a heuristic approach to determine an optimized robot evolution tree. Experiments have shown that our method is able to improve the efficiency of one-to-three transfer of manipulation policy by up to 3.2$\times$ and one-to-six transfer of agile locomotion policy by 2.4$\times$ in terms of simulation cost over the baseline of launching multiple independent one-to-one policy transfers. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_03534 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Meta-Evolve: Continuous Robot Evolution for One-to-many Policy Transfer Liu, Xingyu Pathak, Deepak Zhao, Ding Robotics Artificial Intelligence Machine Learning Neural and Evolutionary Computing We investigate the problem of transferring an expert policy from a source robot to multiple different robots. To solve this problem, we propose a method named $Meta$-$Evolve$ that uses continuous robot evolution to efficiently transfer the policy to each target robot through a set of tree-structured evolutionary robot sequences. The robot evolution tree allows the robot evolution paths to be shared, so our approach can significantly outperform naive one-to-one policy transfer. We present a heuristic approach to determine an optimized robot evolution tree. Experiments have shown that our method is able to improve the efficiency of one-to-three transfer of manipulation policy by up to 3.2$\times$ and one-to-six transfer of agile locomotion policy by 2.4$\times$ in terms of simulation cost over the baseline of launching multiple independent one-to-one policy transfers. |
| title | Meta-Evolve: Continuous Robot Evolution for One-to-many Policy Transfer |
| topic | Robotics Artificial Intelligence Machine Learning Neural and Evolutionary Computing |
| url | https://arxiv.org/abs/2405.03534 |