Riemannian Motion Policy Fusion through Learnable Lyapunov Function Reshaping

Mustafa Mukadam, Ching-An Cheng, Dieter Fox, Byron Boots, Nathan Ratliff
Proceedings of the Conference on Robot Learning, PMLR 100:204-219, 2020.

Abstract

RMPflow is a recently proposed policy-fusion framework based on differential geometry. While RMPflow has demonstrated promising performance, it requires the user to provide sensible subtask policies as Riemannian motion policies (RMPs: a motion policy and an importance matrix function), which can be a difficult design problem in its own right. We propose RMPfusion, a variation of RMPflow, to address this issue. RMPfusion supplements RMPflow with weight functions that can hierarchically reshape the Lyapunov functions of the subtask RMPs according to the current configuration of the robot and environment. This extra flexibility can remedy imperfect subtask RMPs provided by the user, improving the combined policy’s performance. These weight functions can be learned by back-propagation. Moreover, we prove that, under mild restrictions on the weight functions, RMPfusion always yields a globally Lyapunov-stable motion policy. This implies that we can treat RMPfusion as a structured policy class in policy optimization that is guaranteed to generate stable policies, even during the immature phase of learning. We demonstrate these properties of RMPfusion in imitation learning experiments both in simulation and on a real-world robot.

Cite this Paper

BibTeX
@InProceedings{pmlr-v100-mukadam20a, title = {Riemannian Motion Policy Fusion through Learnable Lyapunov Function Reshaping}, author = {Mukadam, Mustafa and Cheng, Ching-An and Fox, Dieter and Boots, Byron and Ratliff, Nathan}, booktitle = {Proceedings of the Conference on Robot Learning}, pages = {204--219}, 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/mukadam20a/mukadam20a.pdf}, url = {https://proceedings.mlr.press/v100/mukadam20a.html}, abstract = {RMPflow is a recently proposed policy-fusion framework based on differential geometry. While RMPflow has demonstrated promising performance, it requires the user to provide sensible subtask policies as Riemannian motion policies (RMPs: a motion policy and an importance matrix function), which can be a difficult design problem in its own right. We propose RMPfusion, a variation of RMPflow, to address this issue. RMPfusion supplements RMPflow with weight functions that can hierarchically reshape the Lyapunov functions of the subtask RMPs according to the current configuration of the robot and environment. This extra flexibility can remedy imperfect subtask RMPs provided by the user, improving the combined policy’s performance. These weight functions can be learned by back-propagation. Moreover, we prove that, under mild restrictions on the weight functions, RMPfusion always yields a globally Lyapunov-stable motion policy. This implies that we can treat RMPfusion as a structured policy class in policy optimization that is guaranteed to generate stable policies, even during the immature phase of learning. We demonstrate these properties of RMPfusion in imitation learning experiments both in simulation and on a real-world robot.} }
Endnote
%0 Conference Paper %T Riemannian Motion Policy Fusion through Learnable Lyapunov Function Reshaping %A Mustafa Mukadam %A Ching-An Cheng %A Dieter Fox %A Byron Boots %A Nathan Ratliff %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-mukadam20a %I PMLR %P 204--219 %U https://proceedings.mlr.press/v100/mukadam20a.html %V 100 %X RMPflow is a recently proposed policy-fusion framework based on differential geometry. While RMPflow has demonstrated promising performance, it requires the user to provide sensible subtask policies as Riemannian motion policies (RMPs: a motion policy and an importance matrix function), which can be a difficult design problem in its own right. We propose RMPfusion, a variation of RMPflow, to address this issue. RMPfusion supplements RMPflow with weight functions that can hierarchically reshape the Lyapunov functions of the subtask RMPs according to the current configuration of the robot and environment. This extra flexibility can remedy imperfect subtask RMPs provided by the user, improving the combined policy’s performance. These weight functions can be learned by back-propagation. Moreover, we prove that, under mild restrictions on the weight functions, RMPfusion always yields a globally Lyapunov-stable motion policy. This implies that we can treat RMPfusion as a structured policy class in policy optimization that is guaranteed to generate stable policies, even during the immature phase of learning. We demonstrate these properties of RMPfusion in imitation learning experiments both in simulation and on a real-world robot.
APA
Mukadam, M., Cheng, C., Fox, D., Boots, B. & Ratliff, N.. (2020). Riemannian Motion Policy Fusion through Learnable Lyapunov Function Reshaping. Proceedings of the Conference on Robot Learning, in Proceedings of Machine Learning Research 100:204-219 Available from https://proceedings.mlr.press/v100/mukadam20a.html.

Related Material