Safe Optimal Control Using Stochastic Barrier Functions and Deep Forward-Backward SDEs

Marcus Pereira, Ziyi Wang, Ioannis Exarchos, Evangelos Theodorou
Proceedings of the 2020 Conference on Robot Learning, PMLR 155:1783-1801, 2021.

Abstract

This paper introduces a new formulation for stochastic optimal control and stochastic dynamic optimization that ensures safety with respect to state and control constraints. The proposed methodology brings together concepts such as Forward-Backward Stochastic Differential Equations, Stochastic Barrier Functions, Differentiable Convex Optimization and Deep Learning. Using the aforementioned concepts, a Neural Network architecture is designed for safe trajectory optimization in which learning can be performed in an end-to-end fashion. Simulations are performed on three systems to show the efficacy of the proposed methodology.

Cite this Paper

BibTeX
@InProceedings{pmlr-v155-pereira21a, title = {Safe Optimal Control Using Stochastic Barrier Functions and Deep Forward-Backward SDEs}, author = {Pereira, Marcus and Wang, Ziyi and Exarchos, Ioannis and Theodorou, Evangelos}, booktitle = {Proceedings of the 2020 Conference on Robot Learning}, pages = {1783--1801}, 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/pereira21a/pereira21a.pdf}, url = {https://proceedings.mlr.press/v155/pereira21a.html}, abstract = {This paper introduces a new formulation for stochastic optimal control and stochastic dynamic optimization that ensures safety with respect to state and control constraints. The proposed methodology brings together concepts such as Forward-Backward Stochastic Differential Equations, Stochastic Barrier Functions, Differentiable Convex Optimization and Deep Learning. Using the aforementioned concepts, a Neural Network architecture is designed for safe trajectory optimization in which learning can be performed in an end-to-end fashion. Simulations are performed on three systems to show the efficacy of the proposed methodology.} }
Endnote
%0 Conference Paper %T Safe Optimal Control Using Stochastic Barrier Functions and Deep Forward-Backward SDEs %A Marcus Pereira %A Ziyi Wang %A Ioannis Exarchos %A Evangelos Theodorou %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-pereira21a %I PMLR %P 1783--1801 %U https://proceedings.mlr.press/v155/pereira21a.html %V 155 %X This paper introduces a new formulation for stochastic optimal control and stochastic dynamic optimization that ensures safety with respect to state and control constraints. The proposed methodology brings together concepts such as Forward-Backward Stochastic Differential Equations, Stochastic Barrier Functions, Differentiable Convex Optimization and Deep Learning. Using the aforementioned concepts, a Neural Network architecture is designed for safe trajectory optimization in which learning can be performed in an end-to-end fashion. Simulations are performed on three systems to show the efficacy of the proposed methodology.
APA
Pereira, M., Wang, Z., Exarchos, I. & Theodorou, E.. (2021). Safe Optimal Control Using Stochastic Barrier Functions and Deep Forward-Backward SDEs. Proceedings of the 2020 Conference on Robot Learning, in Proceedings of Machine Learning Research 155:1783-1801 Available from https://proceedings.mlr.press/v155/pereira21a.html.

Related Material