Learning from demonstration with model-based Gaussian process

Noémie Jaquier, David Ginsbourger, Sylvain Calinon
Proceedings of the Conference on Robot Learning, PMLR 100:247-257, 2020.

Abstract

In learning from demonstrations, it is often desirable to adapt the behavior of the robot as a function of the variability retrieved from human demonstrations and the (un)certainty encoded in different parts of the task. In this paper, we propose a novel multi-output Gaussian process (MOGP) based on Gaussian mixture regression (GMR). The proposed approach encapsulates the variability retrieved from the demonstrations in the covariance of the MOGP. Leveraging the generative nature of GP models, our approach can efficiently modulate trajectories towards new start-, via- or end-points defined by the task. Our framework allows the robot to precisely track via-points while being compliant in regions of high variability. We illustrate the proposed approach in simulated examples and validate it in a real-robot experiment.

Cite this Paper

BibTeX
@InProceedings{pmlr-v100-jaquier20b, title = {Learning from demonstration with model-based Gaussian process}, author = {Jaquier, No\'emie and Ginsbourger, David and Calinon, Sylvain}, booktitle = {Proceedings of the Conference on Robot Learning}, pages = {247--257}, year = {2020}, editor = {Kaelbling, Leslie Pack and Kragic, Danica and Sugiura, Komei}, volume = {100}, series = {Proceedings of Machine Learning Research}, month = {30 Oct--01 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v100/jaquier20b/jaquier20b.pdf}, url = {https://proceedings.mlr.press/v100/jaquier20b.html}, abstract = {In learning from demonstrations, it is often desirable to adapt the behavior of the robot as a function of the variability retrieved from human demonstrations and the (un)certainty encoded in different parts of the task. In this paper, we propose a novel multi-output Gaussian process (MOGP) based on Gaussian mixture regression (GMR). The proposed approach encapsulates the variability retrieved from the demonstrations in the covariance of the MOGP. Leveraging the generative nature of GP models, our approach can efficiently modulate trajectories towards new start-, via- or end-points defined by the task. Our framework allows the robot to precisely track via-points while being compliant in regions of high variability. We illustrate the proposed approach in simulated examples and validate it in a real-robot experiment.} }
Endnote
%0 Conference Paper %T Learning from demonstration with model-based Gaussian process %A Noémie Jaquier %A David Ginsbourger %A Sylvain Calinon %B Proceedings of the Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2020 %E Leslie Pack Kaelbling %E Danica Kragic %E Komei Sugiura %F pmlr-v100-jaquier20b %I PMLR %P 247--257 %U https://proceedings.mlr.press/v100/jaquier20b.html %V 100 %X In learning from demonstrations, it is often desirable to adapt the behavior of the robot as a function of the variability retrieved from human demonstrations and the (un)certainty encoded in different parts of the task. In this paper, we propose a novel multi-output Gaussian process (MOGP) based on Gaussian mixture regression (GMR). The proposed approach encapsulates the variability retrieved from the demonstrations in the covariance of the MOGP. Leveraging the generative nature of GP models, our approach can efficiently modulate trajectories towards new start-, via- or end-points defined by the task. Our framework allows the robot to precisely track via-points while being compliant in regions of high variability. We illustrate the proposed approach in simulated examples and validate it in a real-robot experiment.
APA
Jaquier, N., Ginsbourger, D. & Calinon, S.. (2020). Learning from demonstration with model-based Gaussian process. Proceedings of the Conference on Robot Learning, in Proceedings of Machine Learning Research 100:247-257 Available from https://proceedings.mlr.press/v100/jaquier20b.html.

Related Material