Learning Hierarchical Task Networks with Preferences from Unannotated Demonstrations

Kevin Chen, Nithin Shrivatsav Srikanth, David Kent, Harish Ravichandar, Sonia Chernova
Proceedings of the 2020 Conference on Robot Learning, PMLR 155:1572-1581, 2021.

Abstract

We address the problem of learning Hierarchical Task Networks (HTNs) from unannotated task demonstrations, while retaining action execution preferences present in the demonstration data. We show that the problem of learning a complex HTN structure can be made analogous to the problem of series/parallel reduction of resistor networks, a foundational concept in Electrical Engineering. Based on this finding, we present the CircuitHTN algorithm, which constructs an action graph representing the demonstrations, and then reduces the graph following rules for reducing combination electrical circuits. Evaluation on real-world household kitchen tasks shows that CircuitHTN outperforms prior work in task structure and preference learning, successfully handling large data sets and exhibiting similar action selection preferences as the demonstrations.

Cite this Paper

BibTeX
@InProceedings{pmlr-v155-chen21d, title = {Learning Hierarchical Task Networks with Preferences from Unannotated Demonstrations}, author = {Chen, Kevin and Srikanth, Nithin Shrivatsav and Kent, David and Ravichandar, Harish and Chernova, Sonia}, booktitle = {Proceedings of the 2020 Conference on Robot Learning}, pages = {1572--1581}, year = {2021}, editor = {Kober, Jens and Ramos, Fabio and Tomlin, Claire}, volume = {155}, series = {Proceedings of Machine Learning Research}, month = {16--18 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v155/chen21d/chen21d.pdf}, url = {https://proceedings.mlr.press/v155/chen21d.html}, abstract = {We address the problem of learning Hierarchical Task Networks (HTNs) from unannotated task demonstrations, while retaining action execution preferences present in the demonstration data. We show that the problem of learning a complex HTN structure can be made analogous to the problem of series/parallel reduction of resistor networks, a foundational concept in Electrical Engineering. Based on this finding, we present the CircuitHTN algorithm, which constructs an action graph representing the demonstrations, and then reduces the graph following rules for reducing combination electrical circuits. Evaluation on real-world household kitchen tasks shows that CircuitHTN outperforms prior work in task structure and preference learning, successfully handling large data sets and exhibiting similar action selection preferences as the demonstrations.} }
Endnote
%0 Conference Paper %T Learning Hierarchical Task Networks with Preferences from Unannotated Demonstrations %A Kevin Chen %A Nithin Shrivatsav Srikanth %A David Kent %A Harish Ravichandar %A Sonia Chernova %B Proceedings of the 2020 Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2021 %E Jens Kober %E Fabio Ramos %E Claire Tomlin %F pmlr-v155-chen21d %I PMLR %P 1572--1581 %U https://proceedings.mlr.press/v155/chen21d.html %V 155 %X We address the problem of learning Hierarchical Task Networks (HTNs) from unannotated task demonstrations, while retaining action execution preferences present in the demonstration data. We show that the problem of learning a complex HTN structure can be made analogous to the problem of series/parallel reduction of resistor networks, a foundational concept in Electrical Engineering. Based on this finding, we present the CircuitHTN algorithm, which constructs an action graph representing the demonstrations, and then reduces the graph following rules for reducing combination electrical circuits. Evaluation on real-world household kitchen tasks shows that CircuitHTN outperforms prior work in task structure and preference learning, successfully handling large data sets and exhibiting similar action selection preferences as the demonstrations.
APA
Chen, K., Srikanth, N.S., Kent, D., Ravichandar, H. & Chernova, S.. (2021). Learning Hierarchical Task Networks with Preferences from Unannotated Demonstrations. Proceedings of the 2020 Conference on Robot Learning, in Proceedings of Machine Learning Research 155:1572-1581 Available from https://proceedings.mlr.press/v155/chen21d.html.

Related Material