Hostname: page-component-586b7cd67f-rdxmf Total loading time: 0 Render date: 2024-11-21T12:10:13.789Z Has data issue: false hasContentIssue false
  • English
  • Français

Submit content

Help address this Question with your content Submit Content

How to ensure safety of learning-enabled cyber-physical systems?

Published online by Cambridge University Press:  14 February 2023

Nicola Paoletti  and
Jim Woodcock
Nicola Paoletti*
Affiliation:
Department of Informatics, King’s College London, UK
Jim Woodcock
Affiliation:
Department of Computer Science, University of York, UK
*
Author for correspondence: Nicola Paoletti, Email: [email protected]
Rights & Permissions [Opens in a new window]

Abstract

Type
Question
Information
Research Directions: Cyber-Physical Systems , Volume 1 , 2023 , e2
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2023. Published by Cambridge University Press

Context

Modern cyber-physical systems (CPS) integrate machine learning and deep learning components for a variety of tasks, including sensing, control, anomaly detection and learning process dynamics from data. Formal verification of CPS is paramount to ensure their correct behaviour in many safety-critical application domains (including robotics, avionics, health and automotive). However, traditional CPS verification methods are designed to work with mechanistic CPS models, and hence cannot deal in general with data-driven components. Therefore, how to guarantee the correct behaviour of learning-enabled CPS is still an open question which must be addressed in order to deploy these systems in real-world safety-critical settings.

Within this research question, we welcome contributions about the formal analysis of learning-enabled CPSs. Examples include but are not limited to:

  • Verifying systems with machine-learning components in the loop (including model checking and theorem proving).

  • Formal reasoning about the model’s (epistemic) uncertainty.

  • Learning safe data-driven models for CPS monitoring and control.

  • Dealing with distribution shifts induced by (possibly adversarial) runtime changes of the CPS (e.g., detecting when existing system properties cease to hold and adapting the verification methods to account for such runtime changes).

How to contribute to this Question

If you believe you can contribute to answering this Question with your research outputs, find out how to submit in the Instructions for authors (https://www.cambridge.org/core/journals/research-directions-cyber-physical-systems/information/author-instructions/preparing-your-materials). This journal publishes Results, Analyses, Impact papers and additional content such as preprints and “grey literature”. Questions will be closed when the editors agree that enough has been published to answer the Question so before submitting, check if this is still an active Question. If it is closed, another relevant Question may be currently open, so do review all the open Questions in your field. For any further queries, check the information pages (https://www.cambridge.org/core/journals/research-directions-cyber-physical-systems/information/about-this-journal) or contact this email ([email protected]).

Competing interests

The authors declare none.

References

References

Seshia, SA, Sadigh, D and Sastry, SS (2022) Toward verified artificial intelligence. Communications of the ACM 65, 46–55.CrossRefGoogle Scholar
Huang, X, et al. (2020) A survey of safety and trustworthiness of deep neural networks: Verification, testing, adversarial attack and defence, and interpretability. Computer Science Review 37, 100270.CrossRefGoogle Scholar
Tambon, F, et al. (2022) How to certify machine learning based safety-critical systems? A systematic literature review. Automated Software Engineering 29, 1–74.CrossRefGoogle Scholar
Tran, H-D, Xiang, W and Johnson, TT (2020) Verification approaches for learning-enabled autonomous cyber-physical systems. IEEE Design & Test 39, 24–34.CrossRefGoogle Scholar
Liu, C, et al. (2021) Algorithms for verifying deep neural networks. Foundations and Trends® in Optimization 4, 244–404.CrossRefGoogle Scholar
Li, L, et al. (2020). SoK: Certified robustness for deep neural networks. ArXiv preprint arXiv:2009.04131.Google Scholar
Lauri, M, Hsu, D and Pajarinen, J (2022) Partially observable Markov decision processes in robotics: A survey. IEEE Transactions on Robotics.Google Scholar
Jansen, N, et al. (2020) Safe reinforcement learning using probabilistic shields. In 31st International Conference on Concurrency Theory (CONCUR 2020). Schloss Dagstuhl-Leibniz-Zentrum für Informatik.Google Scholar
Cleaveland, M, et al. (2022) Risk verification of stochastic systems with neural network controllers. Artificial Intelligence 313, 103782.CrossRefGoogle Scholar
Dixit, A, et al. (2022) Adaptive Conformal Prediction for Motion Planning among Dynamic Agents. ArXiv preprint arXiv:2212.00278.Google Scholar
Cairoli, F, Paoletti, N and Bortolussi, L (2022) Conformal Quantitative Predictive Monitoring of STL Requirements for Stochastic Processes. ArXiv preprint arXiv:2211.02375.Google Scholar
Bortolussi, L, et al. (2021) Neural predictive monitoring and a comparison of frequentist and Bayesian approaches. International Journal on Software Tools for Technology Transfer 23, 615–640.CrossRefGoogle Scholar
Akintunde, ME, et al. (2022) Formal verification of neural agents in non-deterministic environments. Autonomous Agents and Multi-Agent Systems 36, 1–36.CrossRefGoogle Scholar
Yan, R, et al. (2022) Strategy Synthesis for Zero-Sum Neuro-Symbolic Concurrent Stochastic Games. ArXiv preprint arXiv:2202.06255.Google Scholar
Bacci, E and Parker, D (2020) Probabilis tic guarantees for safe deep reinforcement learning. In International Conference on Formal Modeling and Analysis of Timed Systems. Springer, Cham.Google Scholar
Wicker, M, et al. (2021) Certification of iterative predictions in Bayesian neural networks. In Uncertainty in Artificial Intelligence. PMLR.Google Scholar
Seshia, SA, Sadigh, D and Sastry, SS (2022) Toward verified artificial intelligence. Communications of the ACM 65, 46–55.CrossRefGoogle Scholar
Huang, X, et al. (2020) A survey of safety and trustworthiness of deep neural networks: Verification, testing, adversarial attack and defence, and interpretability. Computer Science Review 37, 100270.CrossRefGoogle Scholar
Tambon, F, et al. (2022) How to certify machine learning based safety-critical systems? A systematic literature review. Automated Software Engineering 29, 1–74.CrossRefGoogle Scholar
Tran, H-D, Xiang, W and Johnson, TT (2020) Verification approaches for learning-enabled autonomous cyber-physical systems. IEEE Design & Test 39, 24–34.CrossRefGoogle Scholar
Liu, C, et al. (2021) Algorithms for verifying deep neural networks. Foundations and Trends® in Optimization 4, 244–404.CrossRefGoogle Scholar
Li, L, et al. (2020). SoK: Certified robustness for deep neural networks. ArXiv preprint arXiv:2009.04131.Google Scholar
Lauri, M, Hsu, D and Pajarinen, J (2022) Partially observable Markov decision processes in robotics: A survey. IEEE Transactions on Robotics.Google Scholar
Jansen, N, et al. (2020) Safe reinforcement learning using probabilistic shields. In 31st International Conference on Concurrency Theory (CONCUR 2020). Schloss Dagstuhl-Leibniz-Zentrum für Informatik.Google Scholar
Cleaveland, M, et al. (2022) Risk verification of stochastic systems with neural network controllers. Artificial Intelligence 313, 103782.CrossRefGoogle Scholar
Dixit, A, et al. (2022) Adaptive Conformal Prediction for Motion Planning among Dynamic Agents. ArXiv preprint arXiv:2212.00278.Google Scholar
Cairoli, F, Paoletti, N and Bortolussi, L (2022) Conformal Quantitative Predictive Monitoring of STL Requirements for Stochastic Processes. ArXiv preprint arXiv:2211.02375.Google Scholar
Bortolussi, L, et al. (2021) Neural predictive monitoring and a comparison of frequentist and Bayesian approaches. International Journal on Software Tools for Technology Transfer 23, 615–640.CrossRefGoogle Scholar
Akintunde, ME, et al. (2022) Formal verification of neural agents in non-deterministic environments. Autonomous Agents and Multi-Agent Systems 36, 1–36.CrossRefGoogle Scholar
Yan, R, et al. (2022) Strategy Synthesis for Zero-Sum Neuro-Symbolic Concurrent Stochastic Games. ArXiv preprint arXiv:2202.06255.Google Scholar
Bacci, E and Parker, D (2020) Probabilis tic guarantees for safe deep reinforcement learning. In International Conference on Formal Modeling and Analysis of Timed Systems. Springer, Cham.Google Scholar
Wicker, M, et al. (2021) Certification of iterative predictions in Bayesian neural networks. In Uncertainty in Artificial Intelligence. PMLR.Google Scholar