Learning Robot Skills with Temporal Variational Inference

Tanmay Shankar, Abhinav Gupta
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:8624-8633, 2020.

Abstract

In this paper, we address the discovery of robotic options from demonstrations in an unsupervised manner. Specifically, we present a framework to jointly learn low-level control policies and higher-level policies of how to use them from demonstrations of a robot performing various tasks. By representing options as continuous latent variables, we frame the problem of learning these options as latent variable inference. We then present a temporally causal variant of variational inference based on a temporal factorization of trajectory likelihoods, that allows us to infer options in an unsupervised manner. We demonstrate the ability of our framework to learn such options across three robotic demonstration datasets, and provide our code.

Cite this Paper

BibTeX
@InProceedings{pmlr-v119-shankar20b, title = {Learning Robot Skills with Temporal Variational Inference}, author = {Shankar, Tanmay and Gupta, Abhinav}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {8624--8633}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/shankar20b/shankar20b.pdf}, url = {https://proceedings.mlr.press/v119/shankar20b.html}, abstract = {In this paper, we address the discovery of robotic options from demonstrations in an unsupervised manner. Specifically, we present a framework to jointly learn low-level control policies and higher-level policies of how to use them from demonstrations of a robot performing various tasks. By representing options as continuous latent variables, we frame the problem of learning these options as latent variable inference. We then present a temporally causal variant of variational inference based on a temporal factorization of trajectory likelihoods, that allows us to infer options in an unsupervised manner. We demonstrate the ability of our framework to learn such options across three robotic demonstration datasets, and provide our code.} }
Endnote
%0 Conference Paper %T Learning Robot Skills with Temporal Variational Inference %A Tanmay Shankar %A Abhinav Gupta %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-shankar20b %I PMLR %P 8624--8633 %U https://proceedings.mlr.press/v119/shankar20b.html %V 119 %X In this paper, we address the discovery of robotic options from demonstrations in an unsupervised manner. Specifically, we present a framework to jointly learn low-level control policies and higher-level policies of how to use them from demonstrations of a robot performing various tasks. By representing options as continuous latent variables, we frame the problem of learning these options as latent variable inference. We then present a temporally causal variant of variational inference based on a temporal factorization of trajectory likelihoods, that allows us to infer options in an unsupervised manner. We demonstrate the ability of our framework to learn such options across three robotic demonstration datasets, and provide our code.
APA
Shankar, T. & Gupta, A.. (2020). Learning Robot Skills with Temporal Variational Inference. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:8624-8633 Available from https://proceedings.mlr.press/v119/shankar20b.html.

Related Material