Modularity through Attention: Efficient Training and Transfer of Language-Conditioned Policies for Robot Manipulation

Yifan Zhou, Shubham Sonawani, Mariano Phielipp, Simon Stepputtis, Heni Amor
Proceedings of The 6th Conference on Robot Learning, PMLR 205:1684-1695, 2023.

Abstract

Language-conditioned policies allow robots to interpret and execute human instructions. Learning such policies requires a substantial investment with regards to time and compute resources. Still, the resulting controllers are highly device-specific and cannot easily be transferred to a robot with different morphology, capability, appearance or dynamics. In this paper, we propose a sample-efficient approach for training language-conditioned manipulation policies that allows for rapid transfer across different types of robots. By introducing a novel method, namely Hierarchical Modularity, and adopting supervised attention across multiple sub-modules, we bridge the divide between modular and end-to-end learning and enable the reuse of functional building blocks. In both simulated and real world robot manipulation experiments, we demonstrate that our method outperforms the current state-of-the-art methods and can transfer policies across 4 different robots in a sample-efficient manner. Finally, we show that the functionality of learned sub-modules is maintained beyond the training process and can be used to introspect the robot decision-making process.

Cite this Paper

BibTeX
@InProceedings{pmlr-v205-zhou23b, title = {Modularity through Attention: Efficient Training and Transfer of Language-Conditioned Policies for Robot Manipulation}, author = {Zhou, Yifan and Sonawani, Shubham and Phielipp, Mariano and Stepputtis, Simon and Amor, Heni}, booktitle = {Proceedings of The 6th Conference on Robot Learning}, pages = {1684--1695}, year = {2023}, editor = {Liu, Karen and Kulic, Dana and Ichnowski, Jeff}, volume = {205}, series = {Proceedings of Machine Learning Research}, month = {14--18 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v205/zhou23b/zhou23b.pdf}, url = {https://proceedings.mlr.press/v205/zhou23b.html}, abstract = {Language-conditioned policies allow robots to interpret and execute human instructions. Learning such policies requires a substantial investment with regards to time and compute resources. Still, the resulting controllers are highly device-specific and cannot easily be transferred to a robot with different morphology, capability, appearance or dynamics. In this paper, we propose a sample-efficient approach for training language-conditioned manipulation policies that allows for rapid transfer across different types of robots. By introducing a novel method, namely Hierarchical Modularity, and adopting supervised attention across multiple sub-modules, we bridge the divide between modular and end-to-end learning and enable the reuse of functional building blocks. In both simulated and real world robot manipulation experiments, we demonstrate that our method outperforms the current state-of-the-art methods and can transfer policies across 4 different robots in a sample-efficient manner. Finally, we show that the functionality of learned sub-modules is maintained beyond the training process and can be used to introspect the robot decision-making process.} }
Endnote
%0 Conference Paper %T Modularity through Attention: Efficient Training and Transfer of Language-Conditioned Policies for Robot Manipulation %A Yifan Zhou %A Shubham Sonawani %A Mariano Phielipp %A Simon Stepputtis %A Heni Amor %B Proceedings of The 6th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2023 %E Karen Liu %E Dana Kulic %E Jeff Ichnowski %F pmlr-v205-zhou23b %I PMLR %P 1684--1695 %U https://proceedings.mlr.press/v205/zhou23b.html %V 205 %X Language-conditioned policies allow robots to interpret and execute human instructions. Learning such policies requires a substantial investment with regards to time and compute resources. Still, the resulting controllers are highly device-specific and cannot easily be transferred to a robot with different morphology, capability, appearance or dynamics. In this paper, we propose a sample-efficient approach for training language-conditioned manipulation policies that allows for rapid transfer across different types of robots. By introducing a novel method, namely Hierarchical Modularity, and adopting supervised attention across multiple sub-modules, we bridge the divide between modular and end-to-end learning and enable the reuse of functional building blocks. In both simulated and real world robot manipulation experiments, we demonstrate that our method outperforms the current state-of-the-art methods and can transfer policies across 4 different robots in a sample-efficient manner. Finally, we show that the functionality of learned sub-modules is maintained beyond the training process and can be used to introspect the robot decision-making process.
APA
Zhou, Y., Sonawani, S., Phielipp, M., Stepputtis, S. & Amor, H.. (2023). Modularity through Attention: Efficient Training and Transfer of Language-Conditioned Policies for Robot Manipulation. Proceedings of The 6th Conference on Robot Learning, in Proceedings of Machine Learning Research 205:1684-1695 Available from https://proceedings.mlr.press/v205/zhou23b.html.

Related Material