Cross-Domain Transfer via Semantic Skill Imitation

Karl Pertsch, Ruta Desai, Vikash Kumar, Franziska Meier, Joseph J Lim, Dhruv Batra, Akshara Rai
Proceedings of The 6th Conference on Robot Learning, PMLR 205:690-700, 2023.

Abstract

We propose an approach for semantic imitation, which uses demonstrations from a source domain, e.g. human videos, to accelerate reinforcement learning (RL) in a different target domain, e.g. a robotic manipulator in a simulated kitchen. Instead of imitating low-level actions like joint velocities, our approach imitates the sequence of demonstrated semantic skills like "opening the microwave" or "turning on the stove". This allows us to transfer demonstrations across environments (e.g. real-world to simulated kitchen) and agent embodiments (e.g. bimanual human demonstration to robotic arm). We evaluate on three challenging cross-domain learning problems and match the performance of demonstration-accelerated RL approaches that require in-domain demonstrations. In a simulated kitchen environment, our approach learns long-horizon robot manipulation tasks, using less than 3 minutes of human video demonstrations from a real-world kitchen. This enables scaling robot learning via the reuse of demonstrations, e.g. collected as human videos, for learning in any number of target domains.

Cite this Paper

BibTeX
@InProceedings{pmlr-v205-pertsch23a, title = {Cross-Domain Transfer via Semantic Skill Imitation}, author = {Pertsch, Karl and Desai, Ruta and Kumar, Vikash and Meier, Franziska and Lim, Joseph J and Batra, Dhruv and Rai, Akshara}, booktitle = {Proceedings of The 6th Conference on Robot Learning}, pages = {690--700}, 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/pertsch23a/pertsch23a.pdf}, url = {https://proceedings.mlr.press/v205/pertsch23a.html}, abstract = {We propose an approach for semantic imitation, which uses demonstrations from a source domain, e.g. human videos, to accelerate reinforcement learning (RL) in a different target domain, e.g. a robotic manipulator in a simulated kitchen. Instead of imitating low-level actions like joint velocities, our approach imitates the sequence of demonstrated semantic skills like "opening the microwave" or "turning on the stove". This allows us to transfer demonstrations across environments (e.g. real-world to simulated kitchen) and agent embodiments (e.g. bimanual human demonstration to robotic arm). We evaluate on three challenging cross-domain learning problems and match the performance of demonstration-accelerated RL approaches that require in-domain demonstrations. In a simulated kitchen environment, our approach learns long-horizon robot manipulation tasks, using less than 3 minutes of human video demonstrations from a real-world kitchen. This enables scaling robot learning via the reuse of demonstrations, e.g. collected as human videos, for learning in any number of target domains.} }
Endnote
%0 Conference Paper %T Cross-Domain Transfer via Semantic Skill Imitation %A Karl Pertsch %A Ruta Desai %A Vikash Kumar %A Franziska Meier %A Joseph J Lim %A Dhruv Batra %A Akshara Rai %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-pertsch23a %I PMLR %P 690--700 %U https://proceedings.mlr.press/v205/pertsch23a.html %V 205 %X We propose an approach for semantic imitation, which uses demonstrations from a source domain, e.g. human videos, to accelerate reinforcement learning (RL) in a different target domain, e.g. a robotic manipulator in a simulated kitchen. Instead of imitating low-level actions like joint velocities, our approach imitates the sequence of demonstrated semantic skills like "opening the microwave" or "turning on the stove". This allows us to transfer demonstrations across environments (e.g. real-world to simulated kitchen) and agent embodiments (e.g. bimanual human demonstration to robotic arm). We evaluate on three challenging cross-domain learning problems and match the performance of demonstration-accelerated RL approaches that require in-domain demonstrations. In a simulated kitchen environment, our approach learns long-horizon robot manipulation tasks, using less than 3 minutes of human video demonstrations from a real-world kitchen. This enables scaling robot learning via the reuse of demonstrations, e.g. collected as human videos, for learning in any number of target domains.
APA
Pertsch, K., Desai, R., Kumar, V., Meier, F., Lim, J.J., Batra, D. & Rai, A.. (2023). Cross-Domain Transfer via Semantic Skill Imitation. Proceedings of The 6th Conference on Robot Learning, in Proceedings of Machine Learning Research 205:690-700 Available from https://proceedings.mlr.press/v205/pertsch23a.html.

Related Material