Learning Efficient Abstract Planning Models that Choose What to Predict

Nishanth Kumar, Willie McClinton, Rohan Chitnis, Tom Silver, Tomás Lozano-Pérez, Leslie Pack Kaelbling
Proceedings of The 7th Conference on Robot Learning, PMLR 229:2070-2095, 2023.

Abstract

An effective approach to solving long-horizon tasks in robotics domains with continuous state and action spaces is bilevel planning, wherein a high-level search over an abstraction of an environment is used to guide low-level decision-making. Recent work has shown how to enable such bilevel planning by learning abstract models in the form of symbolic operators and neural samplers. In this work, we show that existing symbolic operator learning approaches fall short in many robotics domains where a robot’s actions tend to cause a large number of irrelevant changes in the abstract state. This is primarily because they attempt to learn operators that exactly predict all observed changes in the abstract state. To overcome this issue, we propose to learn operators that ‘choose what to predict’ by only modelling changes necessary for abstract planning to achieve specified goals. Experimentally, we show that our approach learns operators that lead to efficient planning across 10 different hybrid robotics domains, including 4 from the challenging BEHAVIOR-100 benchmark, while generalizing to novel initial states, goals, and objects.

Cite this Paper


BibTeX
@InProceedings{pmlr-v229-kumar23a, title = {Learning Efficient Abstract Planning Models that Choose What to Predict}, author = {Kumar, Nishanth and McClinton, Willie and Chitnis, Rohan and Silver, Tom and Lozano-P\'{e}rez, Tom\'{a}s and Kaelbling, Leslie Pack}, booktitle = {Proceedings of The 7th Conference on Robot Learning}, pages = {2070--2095}, year = {2023}, editor = {Tan, Jie and Toussaint, Marc and Darvish, Kourosh}, volume = {229}, series = {Proceedings of Machine Learning Research}, month = {06--09 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v229/kumar23a/kumar23a.pdf}, url = {https://proceedings.mlr.press/v229/kumar23a.html}, abstract = {An effective approach to solving long-horizon tasks in robotics domains with continuous state and action spaces is bilevel planning, wherein a high-level search over an abstraction of an environment is used to guide low-level decision-making. Recent work has shown how to enable such bilevel planning by learning abstract models in the form of symbolic operators and neural samplers. In this work, we show that existing symbolic operator learning approaches fall short in many robotics domains where a robot’s actions tend to cause a large number of irrelevant changes in the abstract state. This is primarily because they attempt to learn operators that exactly predict all observed changes in the abstract state. To overcome this issue, we propose to learn operators that ‘choose what to predict’ by only modelling changes necessary for abstract planning to achieve specified goals. Experimentally, we show that our approach learns operators that lead to efficient planning across 10 different hybrid robotics domains, including 4 from the challenging BEHAVIOR-100 benchmark, while generalizing to novel initial states, goals, and objects.} }
Endnote
%0 Conference Paper %T Learning Efficient Abstract Planning Models that Choose What to Predict %A Nishanth Kumar %A Willie McClinton %A Rohan Chitnis %A Tom Silver %A Tomás Lozano-Pérez %A Leslie Pack Kaelbling %B Proceedings of The 7th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2023 %E Jie Tan %E Marc Toussaint %E Kourosh Darvish %F pmlr-v229-kumar23a %I PMLR %P 2070--2095 %U https://proceedings.mlr.press/v229/kumar23a.html %V 229 %X An effective approach to solving long-horizon tasks in robotics domains with continuous state and action spaces is bilevel planning, wherein a high-level search over an abstraction of an environment is used to guide low-level decision-making. Recent work has shown how to enable such bilevel planning by learning abstract models in the form of symbolic operators and neural samplers. In this work, we show that existing symbolic operator learning approaches fall short in many robotics domains where a robot’s actions tend to cause a large number of irrelevant changes in the abstract state. This is primarily because they attempt to learn operators that exactly predict all observed changes in the abstract state. To overcome this issue, we propose to learn operators that ‘choose what to predict’ by only modelling changes necessary for abstract planning to achieve specified goals. Experimentally, we show that our approach learns operators that lead to efficient planning across 10 different hybrid robotics domains, including 4 from the challenging BEHAVIOR-100 benchmark, while generalizing to novel initial states, goals, and objects.
APA
Kumar, N., McClinton, W., Chitnis, R., Silver, T., Lozano-Pérez, T. & Kaelbling, L.P.. (2023). Learning Efficient Abstract Planning Models that Choose What to Predict. Proceedings of The 7th Conference on Robot Learning, in Proceedings of Machine Learning Research 229:2070-2095 Available from https://proceedings.mlr.press/v229/kumar23a.html.

Related Material