Learning Equality Constraints for Motion Planning on Manifolds

Giovanni Sutanto, Isabel Rayas Fernández, Peter Englert, Ragesh Kumar Ramachandran, Gaurav Sukhatme
Proceedings of the 2020 Conference on Robot Learning, PMLR 155:2292-2305, 2021.

Abstract

Constrained robot motion planning is a widely used technique to solve complex robot tasks. We consider the problem of learning representations of constraints from demonstrations with a deep neural network, which we call Equality Constraint Manifold Neural Network (ECoMaNN). The key idea is to learn a level-set function of the constraint suitable for integration into a constrained sampling-based motion planner. Learning proceeds by aligning subspaces in the network with subspaces of the data. We combine both learned constraints and analytically described constraints into the planner and use a projection-based strategy to find valid points. We evaluate ECoMaNN on its representation capabilities of constraint manifolds, the impact of its individual loss terms, and the motions produced when incorporated into a planner.

Cite this Paper

BibTeX
@InProceedings{pmlr-v155-sutanto21a, title = {Learning Equality Constraints for Motion Planning on Manifolds}, author = {Sutanto, Giovanni and Fern\'{a}ndez, Isabel Rayas and Englert, Peter and Ramachandran, Ragesh Kumar and Sukhatme, Gaurav}, booktitle = {Proceedings of the 2020 Conference on Robot Learning}, pages = {2292--2305}, 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/sutanto21a/sutanto21a.pdf}, url = {https://proceedings.mlr.press/v155/sutanto21a.html}, abstract = {Constrained robot motion planning is a widely used technique to solve complex robot tasks. We consider the problem of learning representations of constraints from demonstrations with a deep neural network, which we call Equality Constraint Manifold Neural Network (ECoMaNN). The key idea is to learn a level-set function of the constraint suitable for integration into a constrained sampling-based motion planner. Learning proceeds by aligning subspaces in the network with subspaces of the data. We combine both learned constraints and analytically described constraints into the planner and use a projection-based strategy to find valid points. We evaluate ECoMaNN on its representation capabilities of constraint manifolds, the impact of its individual loss terms, and the motions produced when incorporated into a planner.} }
Endnote
%0 Conference Paper %T Learning Equality Constraints for Motion Planning on Manifolds %A Giovanni Sutanto %A Isabel Rayas Fernández %A Peter Englert %A Ragesh Kumar Ramachandran %A Gaurav Sukhatme %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-sutanto21a %I PMLR %P 2292--2305 %U https://proceedings.mlr.press/v155/sutanto21a.html %V 155 %X Constrained robot motion planning is a widely used technique to solve complex robot tasks. We consider the problem of learning representations of constraints from demonstrations with a deep neural network, which we call Equality Constraint Manifold Neural Network (ECoMaNN). The key idea is to learn a level-set function of the constraint suitable for integration into a constrained sampling-based motion planner. Learning proceeds by aligning subspaces in the network with subspaces of the data. We combine both learned constraints and analytically described constraints into the planner and use a projection-based strategy to find valid points. We evaluate ECoMaNN on its representation capabilities of constraint manifolds, the impact of its individual loss terms, and the motions produced when incorporated into a planner.
APA
Sutanto, G., Fernández, I.R., Englert, P., Ramachandran, R.K. & Sukhatme, G.. (2021). Learning Equality Constraints for Motion Planning on Manifolds. Proceedings of the 2020 Conference on Robot Learning, in Proceedings of Machine Learning Research 155:2292-2305 Available from https://proceedings.mlr.press/v155/sutanto21a.html.

Related Material