Eq.Bot: Enhance Robotic Manipulation Learning via Group Equivariant Canonicalization
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_ | 1866915626713350144 |
|---|---|
| author | Deng, Jian Wang, Yuandong Zhu, Yangfu Feng, Tao Wo, Tianyu Shao, Zhenzhou |
| author_facet | Deng, Jian Wang, Yuandong Zhu, Yangfu Feng, Tao Wo, Tianyu Shao, Zhenzhou |
| contents | Robotic manipulation systems are increasingly deployed across diverse domains. Yet existing multi-modal learning frameworks lack inherent guarantees of geometric consistency, struggling to handle spatial transformations such as rotations and translations. While recent works attempt to introduce equivariance through bespoke architectural modifications, these methods suffer from high implementation complexity, computational cost, and poor portability. Inspired by human cognitive processes in spatial reasoning, we propose Eq.Bot, a universal canonicalization framework grounded in SE(2) group equivariant theory for robotic manipulation learning. Our framework transforms observations into a canonical space, applies an existing policy, and maps the resulting actions back to the original space. As a model-agnostic solution, Eq.Bot aims to endow models with spatial equivariance without requiring architectural modifications. Extensive experiments demonstrate the superiority of Eq.Bot under both CNN-based (e.g., CLIPort) and Transformer-based (e.g., OpenVLA-OFT) architectures over existing methods on various robotic manipulation tasks, where the most significant improvement can reach 50.0%. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2511_15194 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Eq.Bot: Enhance Robotic Manipulation Learning via Group Equivariant Canonicalization Deng, Jian Wang, Yuandong Zhu, Yangfu Feng, Tao Wo, Tianyu Shao, Zhenzhou Robotics Artificial Intelligence 68T40 (Primary), 68T07, 93C85, 20C35 (Secondary) Robotic manipulation systems are increasingly deployed across diverse domains. Yet existing multi-modal learning frameworks lack inherent guarantees of geometric consistency, struggling to handle spatial transformations such as rotations and translations. While recent works attempt to introduce equivariance through bespoke architectural modifications, these methods suffer from high implementation complexity, computational cost, and poor portability. Inspired by human cognitive processes in spatial reasoning, we propose Eq.Bot, a universal canonicalization framework grounded in SE(2) group equivariant theory for robotic manipulation learning. Our framework transforms observations into a canonical space, applies an existing policy, and maps the resulting actions back to the original space. As a model-agnostic solution, Eq.Bot aims to endow models with spatial equivariance without requiring architectural modifications. Extensive experiments demonstrate the superiority of Eq.Bot under both CNN-based (e.g., CLIPort) and Transformer-based (e.g., OpenVLA-OFT) architectures over existing methods on various robotic manipulation tasks, where the most significant improvement can reach 50.0%. |
| title | Eq.Bot: Enhance Robotic Manipulation Learning via Group Equivariant Canonicalization |
| topic | Robotics Artificial Intelligence 68T40 (Primary), 68T07, 93C85, 20C35 (Secondary) |
| url | https://arxiv.org/abs/2511.15194 |