mlsanityData to 14 Oct 2026

Safe Reinforcement Learning in Black-Box Environments via Adaptive Shielding

arXiv 2405.18180 ↗First posted 28 May 2024cs.AI cs.LGcited by 6

On the map: Deep reinforcement learning methods → Safe and constrained RL

Abstract (arXiv)

Empowering safe exploration of reinforcement learning (RL) agents during training is a critical challenge towards their deployment in many real-world scenarios. When prior knowledge of the domain or task is unavailable, training RL agents in unknown, black-box environments presents an even greater safety risk. We introduce ADVICE (Adaptive Shielding with a Contrastive Autoencoder), a novel post-shielding technique that distinguishes safe and unsafe features of state-action pairs during training, and uses this knowledge to protect the RL agent from executing actions that yield likely hazardous outcomes. Our comprehensive experimental evaluation against state-of-the-art safe RL exploration techniques shows that ADVICE significantly reduces safety violations (approx 50%) during training, with a competitive outcome reward compared to other techniques.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Placed in Safe and constrained RL, one of 15 directions in Deep reinforcement learning methods. Builds on 30 of its 49 references that are in the map. First posted on arXiv 28 May 2024.

Builds on

30 of 49 references.

References · 30

  1. 2015Continuous control with deep reinforcement learningcited by 3,407
  2. 2016Concrete Problems in AI Safetycited by 1,642
  3. 2019Dota 2 with Large Scale Deep Reinforcement Learningcited by 1,103
  4. 2017Constrained Policy Optimizationcited by 983
  5. 2020Meta-Learning in Neural Networks: A Surveycited by 950
  6. 2020A Review of Uncertainty Quantification in Deep Learning: Techniques, Applications and Challengescited by 819
  7. 2021Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learningcited by 512
  8. 2017Safe Reinforcement Learning via Shieldingcited by 449
  9. 2018Reward Constrained Policy Optimizationcited by 417
  10. 2020Responsive Safety in Reinforcement Learning by PID Lagrangian Methodscited by 316
  11. 2018Safe Exploration in Continuous Action Spacescited by 247
  12. 2020Recovery RL: Safe Reinforcement Learning with Learned Recovery Zonescited by 175
18 more
  1. 2017Emergence of Locomotion Behaviours in Rich Environmentscited by 165
  2. 2023Safety-Gymnasium: A Unified Safe Reinforcement Learning Benchmarkcited by 143
  3. 2020Learning to be Safe: Deep RL with a Safety Criticcited by 116
  4. 2020Conservative Safety Critics for Explorationcited by 113
  5. 2017Uncertainty-Aware Reinforcement Learning for Collision Avoidancecited by 88
  6. 2020Safe Reinforcement Learning via Curriculum Inductioncited by 74
  7. 2022Constrained Variational Policy Optimization for Safe Reinforcement Learningcited by 71
  8. 2021Safe Multi-Agent Reinforcement Learning via Shieldingcited by 61
  9. 2022Safe Reinforcement Learning via Shielding under Partial Observabilitycited by 56
  10. 2023Safe Reinforcement Learning via Probabilistic Logic Shieldscited by 42
  11. 2022Online Shielding for Reinforcement Learningcited by 22
  12. 2022Dynamic Shielding for Reinforcement Learning in Black-Box Environmentscited by 11
  13. 2018Safe Option-Critic: Learning Safety in the Option-Critic Architecturecited by 10
  14. 2021Reinforcement Learning with External Knowledge by using Logical Neural Networkscited by 10
  15. 2023Robust Uncertainty Quantification Using Conformalised Monte Carlo Predictioncited by 10
  16. 2022Automata Learning meets Shieldingcited by 9
  17. 2022Learning a Shield from Catastrophic Action Effects: Never Repeat the Same Mistakecited by 5
  18. 2023Approximate Model-Based Shielding for Safe Reinforcement Learningcited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 5.

6 papers.

  1. 2026Shields to Guarantee Probabilistic Safety in MDPscited by 2
  2. 2026Robust Shielding for Safe Reinforcement Learningcited by 2
  3. 2025Safe But Not Sorry: Reducing Over-Conservatism in Safety Critics via Uncertainty-Aware Modulationcited by 0
  4. 2026Uncertainty-Aware and Temporally Regulated Expert Advice in Reinforcement Learning for Autonomous Drivingcited by 0
  5. 2026Imagine to Ensure Safety in Hierarchical Reinforcement Learningcited by 0
  6. 2026Adaptive Probabilistic Shielding by Learning MDPs for Safe Reinforcement Learningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Safety through Permissibility: Shield Construction for Fast and Safe Reinforcement Learning
  2. 2024A Dynamic Safety Shield for Safe and Efficient Reinforcement Learning of Navigation Tasks
  3. 2026Easy-to-Use Shielding for Reinforcement Learning
  4. 2026Robust Probabilistic Shielding for Safe Offline Reinforcement Learning
  5. 2025Adaptive Shielding for Safe Reinforcement Learning under Hidden-Parameter Dynamics Shifts
  6. 2025Probabilistic Shielding for Safe Reinforcement Learning
  7. 2024Leveraging Approximate Model-based Shielding for Probabilistic Safety Guarantees in Continuous Environments
  8. 2024A Safe Exploration Strategy for Model-free Task Adaptation in Safety-constrained Grid Environments