Learning Hybrid Control Barrier Functions from Data

Lars Lindemann, Haimin Hu, Alexander Robey, Hanwen Zhang, Dimos Dimarogonas, Stephen Tu, Nikolai Matni
Proceedings of the 2020 Conference on Robot Learning, PMLR 155:1351-1370, 2021.

Abstract

Motivated by the lack of systematic tools to obtain safe control laws for hybrid systems, we propose an optimization-based framework for learning certifiably safe control laws from data. In particular, we assume a setting in which the system dynamics are known and in which data exhibiting safe system behavior is available. We propose hybrid control barrier functions for hybrid systems as a means to synthesize safe control inputs. Based on this notion, we present an optimization-based framework to learn such hybrid control barrier functions from data. Importantly, we identify sufficient conditions on the data such that feasibility of the optimization problem ensures correctness of the learned hybrid control barrier functions, and hence the safety of the system. We illustrate our findings in two simulations studies, including a compass gait walker.

Cite this Paper

BibTeX
@InProceedings{pmlr-v155-lindemann21a, title = {Learning Hybrid Control Barrier Functions from Data}, author = {Lindemann, Lars and Hu, Haimin and Robey, Alexander and Zhang, Hanwen and Dimarogonas, Dimos and Tu, Stephen and Matni, Nikolai}, booktitle = {Proceedings of the 2020 Conference on Robot Learning}, pages = {1351--1370}, 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/lindemann21a/lindemann21a.pdf}, url = {https://proceedings.mlr.press/v155/lindemann21a.html}, abstract = {Motivated by the lack of systematic tools to obtain safe control laws for hybrid systems, we propose an optimization-based framework for learning certifiably safe control laws from data. In particular, we assume a setting in which the system dynamics are known and in which data exhibiting safe system behavior is available. We propose hybrid control barrier functions for hybrid systems as a means to synthesize safe control inputs. Based on this notion, we present an optimization-based framework to learn such hybrid control barrier functions from data. Importantly, we identify sufficient conditions on the data such that feasibility of the optimization problem ensures correctness of the learned hybrid control barrier functions, and hence the safety of the system. We illustrate our findings in two simulations studies, including a compass gait walker.} }
Endnote
%0 Conference Paper %T Learning Hybrid Control Barrier Functions from Data %A Lars Lindemann %A Haimin Hu %A Alexander Robey %A Hanwen Zhang %A Dimos Dimarogonas %A Stephen Tu %A Nikolai Matni %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-lindemann21a %I PMLR %P 1351--1370 %U https://proceedings.mlr.press/v155/lindemann21a.html %V 155 %X Motivated by the lack of systematic tools to obtain safe control laws for hybrid systems, we propose an optimization-based framework for learning certifiably safe control laws from data. In particular, we assume a setting in which the system dynamics are known and in which data exhibiting safe system behavior is available. We propose hybrid control barrier functions for hybrid systems as a means to synthesize safe control inputs. Based on this notion, we present an optimization-based framework to learn such hybrid control barrier functions from data. Importantly, we identify sufficient conditions on the data such that feasibility of the optimization problem ensures correctness of the learned hybrid control barrier functions, and hence the safety of the system. We illustrate our findings in two simulations studies, including a compass gait walker.
APA
Lindemann, L., Hu, H., Robey, A., Zhang, H., Dimarogonas, D., Tu, S. & Matni, N.. (2021). Learning Hybrid Control Barrier Functions from Data. Proceedings of the 2020 Conference on Robot Learning, in Proceedings of Machine Learning Research 155:1351-1370 Available from https://proceedings.mlr.press/v155/lindemann21a.html.

Related Material