mlsanityData to 14 Oct 2026

Safe Reinforcement Learning via Shielding

arXiv 1708.08611 ↗First posted 29 Aug 2017cs.LO cs.AI cs.LGcited by 449

Not on the map. Why?

Cited by 135 in Deep reinforcement learning methods, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Reinforcement learning algorithms discover policies that maximize reward, but do not necessarily guarantee safety during learning or execution phases. We introduce a new approach to learn optimal policies while enforcing properties expressed in temporal logic. To this end, given the temporal logic specification that is to be obeyed by the learning system, we propose to synthesize a reactive system called a shield. The shield is introduced in the traditional learning process in two alternative ways, depending on the location at which the shield is implemented. In the first one, the shield acts each time the learning agent is about to make a decision and provides a list of safe actions. In the second way, the shield is introduced after the learning agent. The shield monitors the actions from the learner and corrects them only if the chosen action causes a violation of the specification. We discuss which requirements a shield must meet to preserve the convergence guarantees of the learner. Finally, we demonstrate the versatility of our approach on several challenging reinforcement learning scenarios.

Key facts. Cited by 449 papers in the map (data through 14 Oct 2026). First posted on arXiv 29 Aug 2017.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2020: 3; 2021: 38; 2022: 57; 2023: 65; 2024: 72; 2025: 70; 2026: 144.

449 papers.

Sort:

By area:

  1. 2021Safe Learning in Robotics: From Learning-Based Control to Safe Reinforcement Learningcited by 371
  2. 2023Do the Rewards Justify the Means? Measuring Trade-Offs Between Rewards and Ethical Behavior in the MACHIAVELLI Benchmarkcited by 182
  3. 2022A Review of Safe Reinforcement Learning: Methods, Theory and Applicationscited by 181
  4. 2023A Survey of Safety and Trustworthiness of Large Language Models through the Lens of Verification and Validationcited by 93
  5. 2021A Survey on Interpretable Reinforcement Learningcited by 73
  6. 2022Constrained Variational Policy Optimization for Safe Reinforcement Learningcited by 71
  7. 2021Safe Multi-Agent Reinforcement Learning via Shieldingcited by 61
  8. 2022Safe Reinforcement Learning via Shielding under Partial Observabilitycited by 56
  9. 2022Efficient Learning of Safe Driving Policy via Human-AI Copilot Optimizationcited by 49
  10. 2021Safe Reinforcement Learning Using Advantage-Based Interventioncited by 47
  11. 2024A Survey of Constraint Formulations in Safe Reinforcement Learningcited by 44
  12. 2021Provable Repair of Deep Neural Networkscited by 42
38 more
  1. 2023Safe Reinforcement Learning via Probabilistic Logic Shieldscited by 42
  2. 2021Learning Barrier Certificates: Towards Safe Reinforcement Learning with Zero Training-time Violationscited by 39
  3. 2022On the Robustness of Safe Reinforcement Learning under Observational Perturbationscited by 38
  4. 2025A Survey of Sim-to-Real Methods in RL: Progress, Prospects and Challenges with Foundation Modelscited by 38
  5. 2022Provably Safe Reinforcement Learning: Conceptual Analysis, Survey, and Benchmarkingcited by 34
  6. 2022Verifying Learning-Based Robotic Navigation Systemscited by 33
  7. 2022ISAACS: Iterative Soft Adversarial Actor-Critic for Safetycited by 28
  8. 2023Safe Exploration in Reinforcement Learning: A Generalized Formulation and Algorithmscited by 28
  9. 2025ProbGuard: Proactive Runtime Monitoring for LLM Agent Safety via Probabilistic Predictioncited by 28
  10. 2022Sim-to-Lab-to-Real: Safe Reinforcement Learning with Shielding and Generalization Guaranteescited by 27
  11. 2022Provably Safe Reinforcement Learning via Action Projection using Reachability Analysis and Polynomial Zonotopescited by 26
  12. 2022Learning Control Policies for Stochastic Systems with Reach-avoid Guaranteescited by 23
  13. 2024CaT: Constraints as Terminations for Legged Locomotion Reinforcement Learningcited by 23
  14. 2022Online Shielding for Reinforcement Learningcited by 22
  15. 2022When to Ask for Help: Proactive Interventions in Autonomous Reinforcement Learningcited by 21
  16. 2022SoK: Explainable Machine Learning for Computer Security Applicationscited by 20
  17. 2021Neural Network Repair with Reachability Analysiscited by 19
  18. 2022COOL-MC: A Comprehensive Tool for Reinforcement Learning and Model Checkingcited by 19
  19. 2022Provably Safe Deep Reinforcement Learning for Robotic Manipulation in Human Environmentscited by 18
  20. 2023Constraint-Conditioned Policy Optimization for Versatile Safe Reinforcement Learningcited by 18
  21. 2020Active Task-Inference-Guided Deep Inverse Reinforcement Learningcited by 17
  22. 2021In a Nutshell, the Human Asked for This: Latent Goals for Following Temporal Specificationscited by 17
  23. 2022Safe Reinforcement Learning of Dynamic High-Dimensional Robotic Tasks: Navigation, Manipulation, Interactioncited by 17
  24. 2025CBF-RL: Safety Filtering Reinforcement Learning in Training with Control Barrier Functionscited by 17
  25. 2024Formally Verifying Deep Reinforcement Learning Controllers with Lyapunov Barrier Certificatescited by 16
  26. 2024Hamilton-Jacobi Reachability in Reinforcement Learning: A Surveycited by 16
  27. 2022SafeLight: A Reinforcement Learning Method toward Collision-free Traffic Signal Controlcited by 15
  28. 2024Safe Reinforcement Learning on the Constraint Manifold: Theory and Applicationscited by 15
  29. 2024Dynamic Model Predictive Shielding for Provably Safe Reinforcement Learningcited by 15
  30. 2021Shielding Atari Games with Bounded Presciencecited by 14
  31. 2023Compositional Policy Learning in Stochastic Control Systems with Formal Guaranteescited by 14
  32. 2025Doctor-R1: Mastering Clinical Inquiry with Experiential Agentic Reinforcement Learningcited by 14
  33. 2022Search-Based Testing of Reinforcement Learningcited by 13
  34. 2024Verification-Guided Shielding for Deep Reinforcement Learningcited by 13
  35. 2025A Survey of Safe Reinforcement Learning and Constrained MDPs: A Technical Survey on Single-Agent and Multi-Agent Safetycited by 13
  36. 2025Systems Security Foundations for Agentic Computingcited by 13
  37. 2022Correct-by-Construction Runtime Enforcement in AI -- A Surveycited by 12
  38. 2023Learning to Control and Coordinate Mixed Traffic Through Robot Vehicles at Complex and Unsignalized Intersectionscited by 12

50 most cited of 449 All 449 (JSON)