Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks
In-hand manipulation and grasp adjustment with dexterous robotic hands is a complex problem that not only requires highly coordinated finger movements but also deals with interaction variability. The control problem becomes even more complex when introducing tactile information into the feedback loo...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Published: |
Frontiers Media SA
2020
|
| Online Access: | https://hdl.handle.net/1721.1/128707 |
| _version_ | 1826213921292288000 |
|---|---|
| author | Fernandes Veiga, Filipe Akrour, Riad Peters, Jan |
| author2 | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory |
| author_facet | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Fernandes Veiga, Filipe Akrour, Riad Peters, Jan |
| author_sort | Fernandes Veiga, Filipe |
| collection | MIT |
| description | In-hand manipulation and grasp adjustment with dexterous robotic hands is a complex problem that not only requires highly coordinated finger movements but also deals with interaction variability. The control problem becomes even more complex when introducing tactile information into the feedback loop. Traditional approaches do not consider tactile feedback and attempt to solve the problem either by relying on complex models that are not always readily available or by constraining the problem in order to make it more tractable. In this paper, we propose a hierarchical control approach where a higher level policy is learned through reinforcement learning, while low level controllers ensure grip stability throughout the manipulation action. The low level controllers are independent grip stabilization controllers based on tactile feedback. The independent controllers allow reinforcement learning approaches to explore the manipulation tasks state-action space in a more structured manner. We show that this structure allows learning the unconstrained task with RL methods that cannot learn it in a non-hierarchical setting. The low level controllers also provide an abstraction to the tactile sensors input, allowing transfer to real robot platforms. We show preliminary results of the transfer of policies trained in simulation to the real robot hand. |
| first_indexed | 2024-09-23T15:56:40Z |
| format | Article |
| id | mit-1721.1/128707 |
| institution | Massachusetts Institute of Technology |
| last_indexed | 2024-09-23T15:56:40Z |
| publishDate | 2020 |
| publisher | Frontiers Media SA |
| record_format | dspace |
| spelling | mit-1721.1/1287072022-10-02T05:16:46Z Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks Fernandes Veiga, Filipe Akrour, Riad Peters, Jan Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory In-hand manipulation and grasp adjustment with dexterous robotic hands is a complex problem that not only requires highly coordinated finger movements but also deals with interaction variability. The control problem becomes even more complex when introducing tactile information into the feedback loop. Traditional approaches do not consider tactile feedback and attempt to solve the problem either by relying on complex models that are not always readily available or by constraining the problem in order to make it more tractable. In this paper, we propose a hierarchical control approach where a higher level policy is learned through reinforcement learning, while low level controllers ensure grip stability throughout the manipulation action. The low level controllers are independent grip stabilization controllers based on tactile feedback. The independent controllers allow reinforcement learning approaches to explore the manipulation tasks state-action space in a more structured manner. We show that this structure allows learning the unconstrained task with RL methods that cannot learn it in a non-hierarchical setting. The low level controllers also provide an abstraction to the tactile sensors input, allowing transfer to real robot platforms. We show preliminary results of the transfer of policies trained in simulation to the real robot hand. 2020-12-01T22:11:14Z 2020-12-01T22:11:14Z 2020-11 2019-12 Article http://purl.org/eprint/type/JournalArticle 2296-9144 https://hdl.handle.net/1721.1/128707 Veiga, Filipe et al. "Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks." Frontiers in Robotics and AI 7 (November 2020): 521448 © 2020 Veiga et al. https://doi.org/10.3389/frobt.2020.521448 Frontiers in Robotics and AI Creative Commons Attribution 4.0 International license https://creativecommons.org/licenses/by/4.0/ application/pdf Frontiers Media SA Frontiers |
| spellingShingle | Fernandes Veiga, Filipe Akrour, Riad Peters, Jan Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks |
| title | Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks |
| title_full | Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks |
| title_fullStr | Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks |
| title_full_unstemmed | Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks |
| title_short | Hierarchical Tactile-Based Control Decomposition of Dexterous In-Hand Manipulation Tasks |
| title_sort | hierarchical tactile based control decomposition of dexterous in hand manipulation tasks |
| url | https://hdl.handle.net/1721.1/128707 |
| work_keys_str_mv | AT fernandesveigafilipe hierarchicaltactilebasedcontroldecompositionofdexterousinhandmanipulationtasks AT akrourriad hierarchicaltactilebasedcontroldecompositionofdexterousinhandmanipulationtasks AT petersjan hierarchicaltactilebasedcontroldecompositionofdexterousinhandmanipulationtasks |