MeMo: Meaningful, Modular Controllers via Noise Injection
Robots are often built from standardized assemblies, (e.g. arms, legs, or fingers), but each robot must be trained from scratch to control all the actuators of all the parts together. In this paper we demonstrate a new approach that takes a single robot and its controller as input and produces a set...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Published: |
Massachusetts Institute of Technology
2024
|
| Online Access: | https://hdl.handle.net/1721.1/156334 |
| _version_ | 1826210840365236224 |
|---|---|
| author | Tjandrasuwita, Megan |
| author2 | Solar-Lezama, Armando |
| author_facet | Solar-Lezama, Armando Tjandrasuwita, Megan |
| author_sort | Tjandrasuwita, Megan |
| collection | MIT |
| description | Robots are often built from standardized assemblies, (e.g. arms, legs, or fingers), but each robot must be trained from scratch to control all the actuators of all the parts together. In this paper we demonstrate a new approach that takes a single robot and its controller as input and produces a set of modular controllers for each of these assemblies such that when a new robot is built from the same parts, its control can be quickly learned by reusing the modular controllers. We achieve this with a framework called MeMo which learns (Me)aningful, (Mo)dular controllers. Specifically, we propose a novel modularity objective to learn an appropriate division of labor among the modules. We demonstrate that this objective can be optimized simultaneously with standard behavior cloning loss via noise injection. We benchmark our framework in locomotion and grasping environments on simple to complex robot morphology transfer. We also show that the modules help in task transfer. On both structure and task transfer, MeMo achieves improved training efficiency to graph neural network and Transformer baselines. |
| first_indexed | 2024-09-23T14:56:38Z |
| format | Thesis |
| id | mit-1721.1/156334 |
| institution | Massachusetts Institute of Technology |
| last_indexed | 2024-09-23T14:56:38Z |
| publishDate | 2024 |
| publisher | Massachusetts Institute of Technology |
| record_format | dspace |
| spelling | mit-1721.1/1563342024-08-22T03:15:30Z MeMo: Meaningful, Modular Controllers via Noise Injection Tjandrasuwita, Megan Solar-Lezama, Armando Matusik, Wojciech Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science Robots are often built from standardized assemblies, (e.g. arms, legs, or fingers), but each robot must be trained from scratch to control all the actuators of all the parts together. In this paper we demonstrate a new approach that takes a single robot and its controller as input and produces a set of modular controllers for each of these assemblies such that when a new robot is built from the same parts, its control can be quickly learned by reusing the modular controllers. We achieve this with a framework called MeMo which learns (Me)aningful, (Mo)dular controllers. Specifically, we propose a novel modularity objective to learn an appropriate division of labor among the modules. We demonstrate that this objective can be optimized simultaneously with standard behavior cloning loss via noise injection. We benchmark our framework in locomotion and grasping environments on simple to complex robot morphology transfer. We also show that the modules help in task transfer. On both structure and task transfer, MeMo achieves improved training efficiency to graph neural network and Transformer baselines. S.M. 2024-08-21T18:57:38Z 2024-08-21T18:57:38Z 2024-05 2024-07-10T12:59:59.934Z Thesis https://hdl.handle.net/1721.1/156334 In Copyright - Educational Use Permitted Copyright retained by author(s) https://rightsstatements.org/page/InC-EDU/1.0/ application/pdf Massachusetts Institute of Technology |
| spellingShingle | Tjandrasuwita, Megan MeMo: Meaningful, Modular Controllers via Noise Injection |
| title | MeMo: Meaningful, Modular Controllers via Noise Injection |
| title_full | MeMo: Meaningful, Modular Controllers via Noise Injection |
| title_fullStr | MeMo: Meaningful, Modular Controllers via Noise Injection |
| title_full_unstemmed | MeMo: Meaningful, Modular Controllers via Noise Injection |
| title_short | MeMo: Meaningful, Modular Controllers via Noise Injection |
| title_sort | memo meaningful modular controllers via noise injection |
| url | https://hdl.handle.net/1721.1/156334 |
| work_keys_str_mv | AT tjandrasuwitamegan memomeaningfulmodularcontrollersvianoiseinjection |