mlsanityData to 14 Oct 2026

Provably Safe Reinforcement Learning via Action Projection using Reachability Analysis and Polynomial Zonotopes

arXiv 2210.10691 ↗First posted 19 Oct 2022cs.RO cs.LG cs.SYcited by 26

Not on the map. Why?

Abstract (arXiv)

While reinforcement learning produces very promising results for many applications, its main disadvantage is the lack of safety guarantees, which prevents its use in safety-critical systems. In this work, we address this issue by a safety shield for nonlinear continuous systems that solve reach-avoid tasks. Our safety shield prevents applying potentially unsafe actions from a reinforcement learning agent by projecting the proposed action to the closest safe action. This approach is called action projection and is implemented via mixed-integer optimization. The safety constraints for action projection are obtained by applying parameterized reachability analysis using polynomial zonotopes, which enables to accurately capture the nonlinear effects of the actions on the system. In contrast to other state-of-the-art approaches for action projection, our safety shield can efficiently handle input constraints and dynamic obstacles, eases incorporation of the spatial robot dimensions into the safety constraints, guarantees robust safety despite process noise and measurement errors, and is well suited for high-dimensional systems, as we demonstrate on several challenging benchmark systems.

Key facts. Cited by 26 papers in the map (data through 14 Oct 2026). Builds on 22 of its 61 references that are in the map. First posted on arXiv 19 Oct 2022.

Builds on

22 of 61 references.

References · 22

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2017Constrained Policy Optimizationcited by 983
  3. 2020Deep Reinforcement Learning for Autonomous Driving: A Surveycited by 721
  4. 2017Safe Reinforcement Learning via Shieldingcited by 449
  5. 2020Sim-to-Real Transfer in Deep Reinforcement Learning for Robotics: a Surveycited by 413
  6. 2019End-to-End Safe Reinforcement Learning through Barrier Functions for Safety-Critical Continuous Control Taskscited by 289
  7. 2020Projection-Based Constrained Policy Optimizationcited by 203
  8. 2020A Closer Look at Invalid Action Masking in Policy Gradient Algorithmscited by 171
  9. 2018A predictive safety filter for learning-based control of constrained nonlinear dynamical systemscited by 118
  10. 2017Trial without Error: Towards Safe Reinforcement Learning via Human Interventioncited by 103
  11. 2019Safe Reinforcement Learning on Autonomous Vehiclescited by 50
  12. 2021safe-control-gym: a Unified Benchmark Suite for Safe Learning-based Control and Reinforcement Learning in Roboticscited by 50
10 more
  1. 2020Cautious Reinforcement Learning with Logical Constraintscited by 45
  2. 2020Verifiably Safe Exploration for End-to-End Reinforcement Learningcited by 34
  3. 2019Neural Simplex Architecturecited by 29
  4. 2020Reachability-based Trajectory Safeguard (RTS): A Safe and Fast Reinforcement Learning Safety Layer for Continuous Controlcited by 28
  5. 2022Safe Reinforcement Learning Using Black-Box Reachability Analysiscited by 26
  6. 2022Provably Safe Deep Reinforcement Learning for Robotic Manipulation in Human Environmentscited by 18
  7. 2019Sparse Polynomial Zonotopes: A Novel Set Representation for Reachability Analysiscited by 14
  8. 2020Online Shielding for Stochastic Systemscited by 11
  9. 2022Fully-Automated Verification of Linear Systems Using Inner- and Outer-Approximations of Reachable Setscited by 4
  10. 2019Utilizing Dependencies to Obtain Subsets of Reachable Setscited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 1; 2023: 5; 2024: 7; 2025: 6; 2026: 7.

26 papers.

Sort:

By area:

  1. 2023Iterative Reachability Estimation for Safe Reinforcement Learningcited by 37
  2. 2023Empowering Autonomous Driving with Large Language Models: A Safety Perspectivecited by 34
  3. 2024NavRL: Learning Safe Flight in Dynamic Environmentscited by 26
  4. 2024Safety Filtering While Training: Improving the Performance and Sample Efficiency of Reinforcement Learning Agentscited by 17
  5. 2023A Human-Centered Safe Robot Reinforcement Learning Framework with Interactive Behaviorscited by 15
  6. 2024Safe Reinforcement Learning on the Constraint Manifold: Theory and Applicationscited by 15
  7. 2023POLAR-Express: Efficient and Precise Formal Reachability Analysis of Neural-Network Controlled Systemscited by 12
  8. 2024Verified Safe Reinforcement Learning for Neural Network Dynamic Modelscited by 7
  9. 2024Reinforcement Learning with Adaptive Regularization for Safe Control of Critical Systemscited by 6
  10. 2022Safe Reinforcement Learning using Data-Driven Predictive Controlcited by 5
  11. 2023DeepBern-Nets: Taming the Complexity of Certifying Neural Networks using Bernstein Polynomial Activations and Precise Bound Propagationcited by 5
  12. 2025Provably Optimal Reinforcement Learning under Safety Filteringcited by 3
14 more
  1. 2024Provable Traffic Rule Compliance in Safe Reinforcement Learning on the Open Seacited by 2
  2. 2026BAP-SRL: Bayesian Adaptive Priority Safe Reinforcement Learning for Vehicle Motion Planning at Mixed Traffic Intersectionscited by 2
  3. 2026Shields to Guarantee Probabilistic Safety in MDPscited by 2
  4. 2026X-Safe: Embodiment-Agnostic Safeguarding for Robot Arm Manipulation Policies via Action Maskingcited by 2
  5. 2025Learning Vision-Based Neural Network Controllers with Semi-Probabilistic Safety Guaranteescited by 1
  6. 2025Leveraging Analytic Gradients in Provably Safe Reinforcement Learningcited by 1
  7. 2025SEG-Parking: Towards Safe, Efficient, and Generalizable Autonomous Parking via End-to-End Offline Reinforcement Learningcited by 1
  8. 2024Safety-Aware Multi-Agent Learning for Dynamic Network Bridgingcited by 0
  9. 2025Towards provable probabilistic safety for scalable embodied AI systemscited by 0
  10. 2025Efficient Safety Verification of Autonomous Vehicles with Neural Network Operatorcited by 0
  11. 2026Conformal Reachability for Safe Control in Unknown Environmentscited by 0
  12. 2026Towards Safe Reinforcement Learning with Reduced Conservativeness: A Case Study on Drone Flight Controlcited by 0
  13. 2026Safe Multi-Robot Coordination via VLM-LLM Reasoning and Reachability Analysiscited by 0
  14. 2026Reachability-Informed Reinforcement Learning for Multi-Impulse Interplanetary Transferscited by 0