Learning Complex Manipulation Tasks from Heterogeneous and Unstructured Demonstrations. Figueroa, N. & Billard, A. In Proceedings of Workshop on Synergies between Learning and Interaction. IEEE/RSJ International Conference on Intelligent Robots and Systems, 2017. abstract bibtex Motivated by the current state-of-the-art in Robot Learning from Demonstration (LfD), in this paper, we tackle two central issues in the learning pipeline: namely, dealing with (1) heterogeneity and (2) unstructuredness in demonstrations of complex manipulation tasks. We build upon our previous work on transform-invariant segmentation and action discovery [1], to learn the underlying action sequence of tasks demonstrated in different reference frames or contexts. We then construct and parametrize a multi-phase task-space control architecture, boot-strapped by the segmented data and model parameters learned from the action discovery approach. Successful case studies of the proposed methodology are presented for uni/bi-manual cooking tasks demonstrated through kinesthetic teaching.
@MISC{Figueroa2017-ID988,
author = {Figueroa, N. and Billard, A.},
title = {Learning Complex Manipulation Tasks from Heterogeneous and Unstructured Demonstrations},
howpublished = {In Proceedings of Workshop on Synergies between Learning and Interaction. IEEE/RSJ International Conference on Intelligent Robots and Systems},
year = {2017},
abstract = {Motivated by the current state-of-the-art in Robot Learning from Demonstration (LfD), in this paper, we tackle two central issues in the learning pipeline: namely, dealing with (1) heterogeneity and (2) unstructuredness in demonstrations of complex manipulation tasks. We build upon our previous work on transform-invariant segmentation and action discovery [1], to learn the underlying action sequence of tasks demonstrated in different reference frames or contexts. We then construct and parametrize a multi-phase task-space control architecture, boot-strapped by the segmented data and model parameters learned from the action discovery approach. Successful case studies of the proposed methodology are presented for uni/bi-manual cooking tasks demonstrated through kinesthetic teaching.}
}
Downloads: 0
{"_id":"6vKEvgxhZZB6mMLjA","bibbaseid":"figueroa-billard-learningcomplexmanipulationtasksfromheterogeneousandunstructureddemonstrations-2017","authorIDs":["6weNhrK95SmkxE6z5","Ni9fMNsJ9JCL6HBPA","W8A2wwWx6khdwJjwK","fJM8yxAntdhgi6xkQ","gYhhMYPjY7vbKsTsr","sRkHv5GAEgoTadzLZ"],"author_short":["Figueroa, N.","Billard, A."],"bibdata":{"bibtype":"misc","type":"misc","author":[{"propositions":[],"lastnames":["Figueroa"],"firstnames":["N."],"suffixes":[]},{"propositions":[],"lastnames":["Billard"],"firstnames":["A."],"suffixes":[]}],"title":"Learning Complex Manipulation Tasks from Heterogeneous and Unstructured Demonstrations","howpublished":"In Proceedings of Workshop on Synergies between Learning and Interaction. IEEE/RSJ International Conference on Intelligent Robots and Systems","year":"2017","abstract":"Motivated by the current state-of-the-art in Robot Learning from Demonstration (LfD), in this paper, we tackle two central issues in the learning pipeline: namely, dealing with (1) heterogeneity and (2) unstructuredness in demonstrations of complex manipulation tasks. We build upon our previous work on transform-invariant segmentation and action discovery [1], to learn the underlying action sequence of tasks demonstrated in different reference frames or contexts. We then construct and parametrize a multi-phase task-space control architecture, boot-strapped by the segmented data and model parameters learned from the action discovery approach. Successful case studies of the proposed methodology are presented for uni/bi-manual cooking tasks demonstrated through kinesthetic teaching.","bibtex":"@MISC{Figueroa2017-ID988,\nauthor = {Figueroa, N. and Billard, A.},\ntitle = {Learning Complex Manipulation Tasks from Heterogeneous and Unstructured Demonstrations},\nhowpublished = {In Proceedings of Workshop on Synergies between Learning and Interaction. IEEE/RSJ International Conference on Intelligent Robots and Systems},\nyear = {2017},\nabstract = {Motivated by the current state-of-the-art in Robot Learning from Demonstration (LfD), in this paper, we tackle two central issues in the learning pipeline: namely, dealing with (1) heterogeneity and (2) unstructuredness in demonstrations of complex manipulation tasks. We build upon our previous work on transform-invariant segmentation and action discovery [1], to learn the underlying action sequence of tasks demonstrated in different reference frames or contexts. We then construct and parametrize a multi-phase task-space control architecture, boot-strapped by the segmented data and model parameters learned from the action discovery approach. Successful case studies of the proposed methodology are presented for uni/bi-manual cooking tasks demonstrated through kinesthetic teaching.}\n}\n\n\n","author_short":["Figueroa, N.","Billard, A."],"key":"Figueroa2017-ID988","id":"Figueroa2017-ID988","bibbaseid":"figueroa-billard-learningcomplexmanipulationtasksfromheterogeneousandunstructureddemonstrations-2017","role":"author","urls":{},"metadata":{"authorlinks":{"figueroa, n":"https://nbfigueroa.github.io/","shavit, y":"https://nbfigueroa.github.io/","dong, h":"https://nbfigueroa.github.io/","mirrazavi salehian, s":"https://nbfigueroa.github.io/","billard, a":"https://nbfigueroa.github.io/","zhou, y":"https://nbfigueroa.github.io/"}},"downloads":0},"bibtype":"misc","biburl":"https://www.dropbox.com/s/x2bon6vqlc79ekc/FigueroaPublications.bib?dl=1","creationDate":"2020-11-07T22:53:48.373Z","downloads":0,"keywords":[],"search_terms":["learning","complex","manipulation","tasks","heterogeneous","unstructured","demonstrations","figueroa","billard"],"title":"Learning Complex Manipulation Tasks from Heterogeneous and Unstructured Demonstrations","year":2017,"dataSources":["QscCmn2DGDfW7oiMm","kJNNQTBYdRiDNqEoj"]}