ArrayBot: Reinforcement Learning for Generalizable Distributed Manipulation through Touch
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2023
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917927889928192 |
|---|---|
| author | Xue, Zhengrong Zhang, Han Cheng, Jingwen He, Zhengmao Ju, Yuanchen Lin, Changyi Zhang, Gu Xu, Huazhe |
| author_facet | Xue, Zhengrong Zhang, Han Cheng, Jingwen He, Zhengmao Ju, Yuanchen Lin, Changyi Zhang, Gu Xu, Huazhe |
| contents | We present ArrayBot, a distributed manipulation system consisting of a $16 \times 16$ array of vertically sliding pillars integrated with tactile sensors, which can simultaneously support, perceive, and manipulate the tabletop objects. Towards generalizable distributed manipulation, we leverage reinforcement learning (RL) algorithms for the automatic discovery of control policies. In the face of the massively redundant actions, we propose to reshape the action space by considering the spatially local action patch and the low-frequency actions in the frequency domain. With this reshaped action space, we train RL agents that can relocate diverse objects through tactile observations only. Surprisingly, we find that the discovered policy can not only generalize to unseen object shapes in the simulator but also transfer to the physical robot without any domain randomization. Leveraging the deployed policy, we present abundant real-world manipulation tasks, illustrating the vast potential of RL on ArrayBot for distributed manipulation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2306_16857 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | ArrayBot: Reinforcement Learning for Generalizable Distributed Manipulation through Touch Xue, Zhengrong Zhang, Han Cheng, Jingwen He, Zhengmao Ju, Yuanchen Lin, Changyi Zhang, Gu Xu, Huazhe Robotics Machine Learning We present ArrayBot, a distributed manipulation system consisting of a $16 \times 16$ array of vertically sliding pillars integrated with tactile sensors, which can simultaneously support, perceive, and manipulate the tabletop objects. Towards generalizable distributed manipulation, we leverage reinforcement learning (RL) algorithms for the automatic discovery of control policies. In the face of the massively redundant actions, we propose to reshape the action space by considering the spatially local action patch and the low-frequency actions in the frequency domain. With this reshaped action space, we train RL agents that can relocate diverse objects through tactile observations only. Surprisingly, we find that the discovered policy can not only generalize to unseen object shapes in the simulator but also transfer to the physical robot without any domain randomization. Leveraging the deployed policy, we present abundant real-world manipulation tasks, illustrating the vast potential of RL on ArrayBot for distributed manipulation. |
| title | ArrayBot: Reinforcement Learning for Generalizable Distributed Manipulation through Touch |
| topic | Robotics Machine Learning |
| url | https://arxiv.org/abs/2306.16857 |