Robust Deep Reinforcement Learning through Adversarial Loss
Not on the map. Why?
Abstract (arXiv)
Recent studies have shown that deep reinforcement learning agents are vulnerable to small adversarial perturbations on the agent's inputs, which raises concerns about deploying such agents in the real world. To address this issue, we propose RADIAL-RL, a principled framework to train reinforcement learning agents with improved robustness against lp-norm bounded adversarial attacks. Our framework is compatible with popular deep reinforcement learning algorithms and we demonstrate its performance with deep Q-learning, A3C and PPO. We experiment on three deep RL benchmarks (Atari, MuJoCo and ProcGen) to show the effectiveness of our robust training algorithm. Our RADIAL-RL agents consistently outperform prior methods when tested against attacks of varying strength and are more computationally efficient to train. In addition, we propose a new evaluation method called Greedy Worst-Case Reward (GWC) to measure attack agnostic robustness of deep RL agents. We show that GWC can be evaluated efficiently and is a good estimate of the reward under the worst possible sequence of adversarial attacks. All code used for our experiments is available at https://github.com/tuomaso/radialrlv2.
Key facts. Cited by 73 papers in the map (data through 7 Oct 2026). First posted on arXiv 5 Aug 2020.
Builds on
No reference list yet.
Built on by
Citing papers per year: 2021: 7; 2022: 15; 2023: 19; 2024: 11; 2025: 14; 2026: 7.
73 papers.
Sort:
By area:
- 2022RAMBO-RL: Robust Adversarial Model-Based Offline Reinforcement Learningcited by 115
- 2021Who Is the Strongest Enemy? Towards Optimal and Efficient Evasion Attacks in Deep RLcited by 54
- 2022Efficient Adversarial Training without Attacking: Worst-Case-Aware Robust Reinforcement Learningcited by 42
- 2021CROP: Certifying Robust Policies for Reinforcement Learning through Functional Smoothingcited by 41
- 2022Trustworthy Reinforcement Learning Against Intrinsic Vulnerabilities: Robustness, Safety, and Generalizabilitycited by 32
- 2022COPA: Certifying Robust Policies for Offline Reinforcement Learning against Poisoning Attackscited by 25
- 2023Adversarial Robust Deep Reinforcement Learning Requires Redefining Robustnesscited by 18
- 2022RoMFAC: A robust mean-field actor-critic reinforcement learning against adversarial perturbations on statescited by 16
- 2024RAT: Adversarial Attacks on Deep Reinforcement Agents for Targeted Behaviorscited by 16
- 2022Robust Deep Reinforcement Learning through Bootstrapped Opportunistic Curriculumcited by 15
- 2023Robust multi-agent coordination via evolutionary generation of auxiliary adversarial attackerscited by 15
- 2021Learning a subspace of policies for online adaptation in Reinforcement Learningcited by 14
38 more
- 2023Games for Artificial Intelligence Research: A Review and Perspectivescited by 14
- 2021Understanding Adversarial Attacks on Observations in Deep Reinforcement Learningcited by 13
- 2022Illusory Attacks: Information-Theoretic Detectability Matters in Adversarial Attackscited by 12
- 2022Distributionally Adaptive Meta Reinforcement Learningcited by 11
- 2021Resilient Machine Learning for Networked Cyber Physical Systems: A Survey for Machine Learning Security to Securing Machine Learning for CPScited by 10
- 2025Rethinking Adversarial Attacks in Reinforcement Learning from Policy Distribution Perspectivecited by 10
- 2021Real-time Adversarial Perturbations against Deep Reinforcement Learning Policies: Attacks and Defensescited by 9
- 2023Adjustable Robust Reinforcement Learning for Online 3D Bin Packingcited by 9
- 2023Robust Adversarial Reinforcement Learning via Bounded Rationality Curriculacited by 9
- 2024Beyond Worst-case Attacks: Robust RL with Adaptive Defense via Non-dominated Policiescited by 8
- 2022Certifiably Robust Policy Learning against Adversarial Communication in Multi-agent Systemscited by 7
- 2022Robust Reinforcement Learning in Continuous Control Tasks with Uncertainty Set Regularizationcited by 7
- 2023Rethinking Adversarial Policies: A Generalized Attack Formulation and Provable Defense in RLcited by 7
- 2023Improve Robustness of Reinforcement Learning against Observation Perturbations via l_∞ Lipschitz Policy Networkscited by 7
- 2025State-Aware Perturbation Optimization for Robust Deep Reinforcement Learningcited by 7
- 2025Robust Deep Reinforcement Learning in Robotics via Adaptive Gradient-Masked Adversarial Attackscited by 7
- 2023Communication-Robust Multi-Agent Learning by Adaptable Auxiliary Multi-Agent Adversary Generationcited by 6
- 2021How Private Is Your RL Policy? An Inverse RL Based Analysis Frameworkcited by 5
- 2024Towards Optimal Adversarial Robust Q-learning with Bellman Infinity-errorcited by 4
- 2024On Robust Reinforcement Learning with Lipschitz-Bounded Policy Networkscited by 4
- 2024Breaking the Barrier: Enhanced Utility and Robustness in Smoothed DRL Agentscited by 4
- 2024Mitigating Adversarial Perturbations for Deep Reinforcement Learning via Vector Quantizationcited by 4
- 2023SoK: Adversarial Machine Learning Attacks and Defences in Multi-Agent Reinforcement Learningcited by 3
- 2023Regret-Based Defense in Adversarial Reinforcement Learningcited by 3
- 2023Toward Evaluating Robustness of Reinforcement Learning with Adversarial Policycited by 3
- 2023Robust Multi-agent Communication via Multi-view Message Certificationcited by 3
- 2023Enhancing the Robustness of QMIX against State-adversarial Attackscited by 3
- 2025Advancing Robustness in Deep Reinforcement Learning with an Ensemble Defense Approachcited by 3
- 2022Review of Metrics to Measure the Stability, Robustness and Resilience of Reinforcement Learningcited by 2
- 2023Certifiably Robust Reinforcement Learning through Model-Based Abstract Interpretationcited by 2
- 2024Belief-Enriched Pessimistic Q-Learning against Adversarial State Perturbationscited by 2
- 2024Robust Deep Reinforcement Learning against Adversarial Behavior Manipulationcited by 2
- 2024On Minimizing Adversarial Counterfactual Error in Adversarial RLcited by 2
- 2025Model-Based Offline Reinforcement Learning with Adversarial Data Augmentationcited by 2
- 2025Enhancing Security in Deep Reinforcement Learning: A Comprehensive Survey on Adversarial Attacks and Defensescited by 2
- 2026Robust Adversarial Policy Optimization Under Dynamics Uncertaintycited by 2
- 2022Red Teaming with Mind Reading: White-Box Adversarial Policies Against RL Agentscited by 1
- 2022Security of Deep Reinforcement Learning for Autonomous Driving: A Surveycited by 1
50 most cited of 73 All 73 (JSON)
- 2026A Unified and Constrained View of Regularization-Based Robust Reinforcement Learning11 Sep 2026
- 2026Robust Policy Optimization via Adversarial Importance Sampling11 Sep 2026
- 2026CIVA: Critic-Induced Value-Subspace Attacks on Visual World-Model Agents21 Aug 2026
- 2026Robust Adversarial Policy Optimization Under Dynamics Uncertainty13 Apr 2026
- 2026Optimizing Reinforcement Learning Training over Digital Twin Enabled Multi-fidelity Networks9 Mar 2026
- 2026Advantage-based Temporal Attack in Reinforcement Learning23 Feb 2026
- 2026TCRL: Temporal-Coupled Adversarial Training for Robust Constrained Reinforcement Learning in Worst-Case Scenarios13 Feb 2026
- 2025On the Tension Between Optimality and Adversarial Robustness in Policy Optimization1 Dec 2025
- 2025Enhancing Security in Deep Reinforcement Learning: A Comprehensive Survey on Adversarial Attacks and Defenses23 Oct 2025
- 2025Advancing Robustness in Deep Reinforcement Learning with an Ensemble Defense Approach22 Jul 2025
- 2025LyAm: Robust Non-Convex Optimization for Stable Learning in Noisy Environments15 Jul 2025
- 2025Advancing Text-to-3D Generation with Linearized Lookahead Variational Score Distillation13 Jul 2025
38 more
- 2025Action Robust Reinforcement Learning via Optimal Adversary Aware Policy Optimization4 Jul 2025
- 2025Off-Policy Actor-Critic for Adversarial Observation Robustness: Virtual Alternative Training via Symmetric Policy Evaluation20 Jun 2025
- 2025Towards Robust Deep Reinforcement Learning against Environmental State Perturbation10 Jun 2025
- 2025Neural Fidelity Calibration for Informative Sim-to-Real Adaptation11 Apr 2025
- 2025Robust Deep Reinforcement Learning in Robotics via Adaptive Gradient-Masked Adversarial Attacks26 Mar 2025
- 2025State-Aware Perturbation Optimization for Robust Deep Reinforcement Learning26 Mar 2025
- 2025Model-Based Offline Reinforcement Learning with Adversarial Data Augmentation26 Mar 2025
- 2025A Dual-Agent Adversarial Framework for Robust Generalization in Deep Reinforcement Learning29 Jan 2025
- 2025Rethinking Adversarial Attacks in Reinforcement Learning from Policy Distribution Perspective7 Jan 2025
- 2024RAT: Adversarial Attacks on Deep Reinforcement Agents for Targeted Behaviors14 Dec 2024
- 2024Mitigating Adversarial Perturbations for Deep Reinforcement Learning via Vector Quantization4 Oct 2024
- 2024Robust off-policy Reinforcement Learning via Soft Constrained Adversary31 Aug 2024
- 2024On the Perturbed States for Transformed Input-robust Reinforcement Learning31 Jul 2024
- 2024Breaking the Barrier: Enhanced Utility and Robustness in Smoothed DRL Agents26 Jun 2024
- 2024On Minimizing Adversarial Counterfactual Error in Adversarial RL7 Jun 2024
- 2024Robust Deep Reinforcement Learning against Adversarial Behavior Manipulation6 Jun 2024
- 2024On Robust Reinforcement Learning with Lipschitz-Bounded Policy Networks19 May 2024
- 2024Belief-Enriched Pessimistic Q-Learning against Adversarial State Perturbations6 Mar 2024
- 2024Beyond Worst-case Attacks: Robust RL with Adaptive Defense via Non-dominated Policies20 Feb 2024
- 2024Towards Optimal Adversarial Robust Q-learning with Bellman Infinity-error3 Feb 2024
- 2023Robustness Verification of Deep Reinforcement Learning Based Control Systems using Reward Martingales15 Dec 2023
- 2023Improve Robustness of Reinforcement Learning against Observation Perturbations via l_∞ Lipschitz Policy Networks14 Dec 2023
- 2023Robust Adversarial Reinforcement Learning via Bounded Rationality Curricula3 Nov 2023
- 2023DSAC-C: Constrained Maximum Entropy for Robust Discrete Soft-Actor Critic26 Oct 2023
- 2023Adjustable Robust Reinforcement Learning for Online 3D Bin Packing6 Oct 2023
- 2023FLARE: Fingerprinting Deep Reinforcement Learning Agents using Universal Adversarial Masks27 Jul 2023
- 2023Game-Theoretic Robust Reinforcement Learning Handles Temporally-Coupled Perturbations22 Jul 2023
- 2023Enhancing the Robustness of QMIX against State-adversarial Attacks3 Jul 2023
- 2023Rethinking Adversarial Policies: A Generalized Attack Formulation and Provable Defense in RL27 May 2023
- 2023Robust multi-agent coordination via evolutionary generation of auxiliary adversarial attackers10 May 2023
- 2023Communication-Robust Multi-Agent Learning by Adaptable Auxiliary Multi-Agent Adversary Generation9 May 2023
- 2023Robust Multi-agent Communication via Multi-view Message Certification7 May 2023
- 2023Toward Evaluating Robustness of Reinforcement Learning with Adversarial Policy4 May 2023
- 2023Games for Artificial Intelligence Research: A Review and Perspectives26 Apr 2023
- 2023Zero-shot Transfer Learning of Driving Policy via Socially Adversarial Traffic Flow25 Apr 2023
- 2023Regret-Based Defense in Adversarial Reinforcement Learning14 Feb 2023
- 2023Certifiably Robust Reinforcement Learning through Model-Based Abstract Interpretation26 Jan 2023
- 2023Adversarial Robust Deep Reinforcement Learning Requires Redefining Robustness17 Jan 2023
50 newest of 73 All 73 (JSON)
- 2021Who Is the Strongest Enemy? Towards Optimal and Efficient Evasion Attacks in Deep RLcited by 54
- 2022Efficient Adversarial Training without Attacking: Worst-Case-Aware Robust Reinforcement Learningcited by 42
- 2021CROP: Certifying Robust Policies for Reinforcement Learning through Functional Smoothingcited by 41
- 2022Trustworthy Reinforcement Learning Against Intrinsic Vulnerabilities: Robustness, Safety, and Generalizabilitycited by 32
- 2023Adversarial Robust Deep Reinforcement Learning Requires Redefining Robustnesscited by 18
- 2022RoMFAC: A robust mean-field actor-critic reinforcement learning against adversarial perturbations on statescited by 16
- 2024RAT: Adversarial Attacks on Deep Reinforcement Agents for Targeted Behaviorscited by 16
- 2022Robust Deep Reinforcement Learning through Bootstrapped Opportunistic Curriculumcited by 15
- 2023Robust multi-agent coordination via evolutionary generation of auxiliary adversarial attackerscited by 15
- 2021Learning a subspace of policies for online adaptation in Reinforcement Learningcited by 14
- 2023Games for Artificial Intelligence Research: A Review and Perspectivescited by 14
- 2021Understanding Adversarial Attacks on Observations in Deep Reinforcement Learningcited by 13
12 most cited of 64 All 64 (JSON)
- 2026A Unified and Constrained View of Regularization-Based Robust Reinforcement Learning11 Sep 2026
- 2026Robust Policy Optimization via Adversarial Importance Sampling11 Sep 2026
- 2026Robust Adversarial Policy Optimization Under Dynamics Uncertainty13 Apr 2026
- 2026Optimizing Reinforcement Learning Training over Digital Twin Enabled Multi-fidelity Networks9 Mar 2026
- 2026Advantage-based Temporal Attack in Reinforcement Learning23 Feb 2026
- 2026TCRL: Temporal-Coupled Adversarial Training for Robust Constrained Reinforcement Learning in Worst-Case Scenarios13 Feb 2026
- 2025On the Tension Between Optimality and Adversarial Robustness in Policy Optimization1 Dec 2025
- 2025Enhancing Security in Deep Reinforcement Learning: A Comprehensive Survey on Adversarial Attacks and Defenses23 Oct 2025
- 2025Advancing Robustness in Deep Reinforcement Learning with an Ensemble Defense Approach22 Jul 2025
- 2025LyAm: Robust Non-Convex Optimization for Stable Learning in Noisy Environments15 Jul 2025
- 2025Action Robust Reinforcement Learning via Optimal Adversary Aware Policy Optimization4 Jul 2025
- 2025Off-Policy Actor-Critic for Adversarial Observation Robustness: Virtual Alternative Training via Symmetric Policy Evaluation20 Jun 2025
12 newest of 64 All 64 (JSON)