Data-driven anytime algorithms for motion planning with safety guarantees

Devesh K. Jha, Minghui Zhu, Yebin Wang, Asok Ray

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

This paper presents a learning-based (i.e., data-driven) approach to motion planning of robotic systems. This is motivated by controller synthesis problems for safety critical systems where an accurate estimate of the uncertainties (e.g., unmodeled dynamics, disturbance) can improve the performance of the system. The state-space of the system is built by sampling from the state-set as well as the input set of the underlying system. The robust adaptive motion planning problem is modeled as a learning-based approach evasion differential game, where a machine-learning algorithm is used to update the statistical estimates of the uncertainties from system observations. The system begins with a conservative estimate of the uncertainty set to ensure safety of the underlying system and we relax the robustness constraints as we get better estimates of the unmodeled uncertainty. The estimates from the machine learning algorithm are used to refine the estimates of the controller in an anytime fashion. We show that the values for the game converges to the optimal values with known disturbance given the statistical estimates on the uncertainty converges. Using confidence intervals for the unmodeled disturbance estimated by the machine learning estimator during the transient learning phase, we are able to guarantee safety of the robotic system with the proposed algorithms during transience.

Original languageEnglish (US)
Title of host publication2016 American Control Conference, ACC 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages5716-5721
Number of pages6
ISBN (Electronic)9781467386821
DOIs
StatePublished - Jul 28 2016
Event2016 American Control Conference, ACC 2016 - Boston, United States
Duration: Jul 6 2016Jul 8 2016

Publication series

NameProceedings of the American Control Conference
Volume2016-July
ISSN (Print)0743-1619

Other

Other2016 American Control Conference, ACC 2016
CountryUnited States
CityBoston
Period7/6/167/8/16

Fingerprint

Motion planning
Learning systems
Learning algorithms
Robotics
Controllers
Uncertainty
Sampling

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Cite this

Jha, D. K., Zhu, M., Wang, Y., & Ray, A. (2016). Data-driven anytime algorithms for motion planning with safety guarantees. In 2016 American Control Conference, ACC 2016 (pp. 5716-5721). [7526565] (Proceedings of the American Control Conference; Vol. 2016-July). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ACC.2016.7526565
Jha, Devesh K. ; Zhu, Minghui ; Wang, Yebin ; Ray, Asok. / Data-driven anytime algorithms for motion planning with safety guarantees. 2016 American Control Conference, ACC 2016. Institute of Electrical and Electronics Engineers Inc., 2016. pp. 5716-5721 (Proceedings of the American Control Conference).
@inproceedings{3c75b42b64f947388a5e5cd6293d0156,
title = "Data-driven anytime algorithms for motion planning with safety guarantees",
abstract = "This paper presents a learning-based (i.e., data-driven) approach to motion planning of robotic systems. This is motivated by controller synthesis problems for safety critical systems where an accurate estimate of the uncertainties (e.g., unmodeled dynamics, disturbance) can improve the performance of the system. The state-space of the system is built by sampling from the state-set as well as the input set of the underlying system. The robust adaptive motion planning problem is modeled as a learning-based approach evasion differential game, where a machine-learning algorithm is used to update the statistical estimates of the uncertainties from system observations. The system begins with a conservative estimate of the uncertainty set to ensure safety of the underlying system and we relax the robustness constraints as we get better estimates of the unmodeled uncertainty. The estimates from the machine learning algorithm are used to refine the estimates of the controller in an anytime fashion. We show that the values for the game converges to the optimal values with known disturbance given the statistical estimates on the uncertainty converges. Using confidence intervals for the unmodeled disturbance estimated by the machine learning estimator during the transient learning phase, we are able to guarantee safety of the robotic system with the proposed algorithms during transience.",
author = "Jha, {Devesh K.} and Minghui Zhu and Yebin Wang and Asok Ray",
year = "2016",
month = "7",
day = "28",
doi = "10.1109/ACC.2016.7526565",
language = "English (US)",
series = "Proceedings of the American Control Conference",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "5716--5721",
booktitle = "2016 American Control Conference, ACC 2016",
address = "United States",

}

Jha, DK, Zhu, M, Wang, Y & Ray, A 2016, Data-driven anytime algorithms for motion planning with safety guarantees. in 2016 American Control Conference, ACC 2016., 7526565, Proceedings of the American Control Conference, vol. 2016-July, Institute of Electrical and Electronics Engineers Inc., pp. 5716-5721, 2016 American Control Conference, ACC 2016, Boston, United States, 7/6/16. https://doi.org/10.1109/ACC.2016.7526565

Data-driven anytime algorithms for motion planning with safety guarantees. / Jha, Devesh K.; Zhu, Minghui; Wang, Yebin; Ray, Asok.

2016 American Control Conference, ACC 2016. Institute of Electrical and Electronics Engineers Inc., 2016. p. 5716-5721 7526565 (Proceedings of the American Control Conference; Vol. 2016-July).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Data-driven anytime algorithms for motion planning with safety guarantees

AU - Jha, Devesh K.

AU - Zhu, Minghui

AU - Wang, Yebin

AU - Ray, Asok

PY - 2016/7/28

Y1 - 2016/7/28

N2 - This paper presents a learning-based (i.e., data-driven) approach to motion planning of robotic systems. This is motivated by controller synthesis problems for safety critical systems where an accurate estimate of the uncertainties (e.g., unmodeled dynamics, disturbance) can improve the performance of the system. The state-space of the system is built by sampling from the state-set as well as the input set of the underlying system. The robust adaptive motion planning problem is modeled as a learning-based approach evasion differential game, where a machine-learning algorithm is used to update the statistical estimates of the uncertainties from system observations. The system begins with a conservative estimate of the uncertainty set to ensure safety of the underlying system and we relax the robustness constraints as we get better estimates of the unmodeled uncertainty. The estimates from the machine learning algorithm are used to refine the estimates of the controller in an anytime fashion. We show that the values for the game converges to the optimal values with known disturbance given the statistical estimates on the uncertainty converges. Using confidence intervals for the unmodeled disturbance estimated by the machine learning estimator during the transient learning phase, we are able to guarantee safety of the robotic system with the proposed algorithms during transience.

AB - This paper presents a learning-based (i.e., data-driven) approach to motion planning of robotic systems. This is motivated by controller synthesis problems for safety critical systems where an accurate estimate of the uncertainties (e.g., unmodeled dynamics, disturbance) can improve the performance of the system. The state-space of the system is built by sampling from the state-set as well as the input set of the underlying system. The robust adaptive motion planning problem is modeled as a learning-based approach evasion differential game, where a machine-learning algorithm is used to update the statistical estimates of the uncertainties from system observations. The system begins with a conservative estimate of the uncertainty set to ensure safety of the underlying system and we relax the robustness constraints as we get better estimates of the unmodeled uncertainty. The estimates from the machine learning algorithm are used to refine the estimates of the controller in an anytime fashion. We show that the values for the game converges to the optimal values with known disturbance given the statistical estimates on the uncertainty converges. Using confidence intervals for the unmodeled disturbance estimated by the machine learning estimator during the transient learning phase, we are able to guarantee safety of the robotic system with the proposed algorithms during transience.

UR - http://www.scopus.com/inward/record.url?scp=84992053062&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84992053062&partnerID=8YFLogxK

U2 - 10.1109/ACC.2016.7526565

DO - 10.1109/ACC.2016.7526565

M3 - Conference contribution

AN - SCOPUS:84992053062

T3 - Proceedings of the American Control Conference

SP - 5716

EP - 5721

BT - 2016 American Control Conference, ACC 2016

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Jha DK, Zhu M, Wang Y, Ray A. Data-driven anytime algorithms for motion planning with safety guarantees. In 2016 American Control Conference, ACC 2016. Institute of Electrical and Electronics Engineers Inc. 2016. p. 5716-5721. 7526565. (Proceedings of the American Control Conference). https://doi.org/10.1109/ACC.2016.7526565