Learning to Compose Hierarchical Object-Centric Controllers for Robotic Manipulation

Mohit Sharma, Jacky Liang, Jialiang Zhao, Alex Lagrassa, Oliver Kroemer
Proceedings of the 2020 Conference on Robot Learning, PMLR 155:822-844, 2021.

Abstract

Manipulation tasks can often be decomposed into multiple subtasks performed in parallel, e.g., sliding an object to a goal pose while maintaining contact with a table. Individual subtasks can be achieved by task-axis controllers defined relative to the objects being manipulated, and a set of object-centric controllers can be combined in an hierarchy. In prior works, such combinations are defined manually or learned from demonstrations. By contrast, we propose using reinforcement learning to dynamically compose hierarchical object-centric controllers for manipulation tasks. Experiments in both simulation and real world show how the proposed approach leads to improved sample efficiency, zero-shot generalization to novel test environments, and simulation-to-reality transfer without fine-tuning.

Cite this Paper

BibTeX
@InProceedings{pmlr-v155-sharma21a, title = {Learning to Compose Hierarchical Object-Centric Controllers for Robotic Manipulation}, author = {Sharma, Mohit and Liang, Jacky and Zhao, Jialiang and Lagrassa, Alex and Kroemer, Oliver}, booktitle = {Proceedings of the 2020 Conference on Robot Learning}, pages = {822--844}, year = {2021}, editor = {Kober, Jens and Ramos, Fabio and Tomlin, Claire}, volume = {155}, series = {Proceedings of Machine Learning Research}, month = {16--18 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v155/sharma21a/sharma21a.pdf}, url = {https://proceedings.mlr.press/v155/sharma21a.html}, abstract = {Manipulation tasks can often be decomposed into multiple subtasks performed in parallel, e.g., sliding an object to a goal pose while maintaining contact with a table. Individual subtasks can be achieved by task-axis controllers defined relative to the objects being manipulated, and a set of object-centric controllers can be combined in an hierarchy. In prior works, such combinations are defined manually or learned from demonstrations. By contrast, we propose using reinforcement learning to dynamically compose hierarchical object-centric controllers for manipulation tasks. Experiments in both simulation and real world show how the proposed approach leads to improved sample efficiency, zero-shot generalization to novel test environments, and simulation-to-reality transfer without fine-tuning.} }
Endnote
%0 Conference Paper %T Learning to Compose Hierarchical Object-Centric Controllers for Robotic Manipulation %A Mohit Sharma %A Jacky Liang %A Jialiang Zhao %A Alex Lagrassa %A Oliver Kroemer %B Proceedings of the 2020 Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2021 %E Jens Kober %E Fabio Ramos %E Claire Tomlin %F pmlr-v155-sharma21a %I PMLR %P 822--844 %U https://proceedings.mlr.press/v155/sharma21a.html %V 155 %X Manipulation tasks can often be decomposed into multiple subtasks performed in parallel, e.g., sliding an object to a goal pose while maintaining contact with a table. Individual subtasks can be achieved by task-axis controllers defined relative to the objects being manipulated, and a set of object-centric controllers can be combined in an hierarchy. In prior works, such combinations are defined manually or learned from demonstrations. By contrast, we propose using reinforcement learning to dynamically compose hierarchical object-centric controllers for manipulation tasks. Experiments in both simulation and real world show how the proposed approach leads to improved sample efficiency, zero-shot generalization to novel test environments, and simulation-to-reality transfer without fine-tuning.
APA
Sharma, M., Liang, J., Zhao, J., Lagrassa, A. & Kroemer, O.. (2021). Learning to Compose Hierarchical Object-Centric Controllers for Robotic Manipulation. Proceedings of the 2020 Conference on Robot Learning, in Proceedings of Machine Learning Research 155:822-844 Available from https://proceedings.mlr.press/v155/sharma21a.html.

Related Material