mlsanityData to 14 Oct 2026

Solving Parameter-Robust Avoid Problems with Unknown Feasibility using Reinforcement Learning

arXiv 2602.15817 ↗First posted 17 Feb 2026cs.LG cs.RO math.OCcited by 0

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

Abstract (arXiv)

Recent advances in deep reinforcement learning (RL) have achieved strong results on high-dimensional control tasks, but applying RL to reachability problems raises a fundamental mismatch: reachability seeks to maximize the set of states from which a system remains safe indefinitely, while RL optimizes expected returns over a user-specified distribution. This mismatch can result in policies that perform poorly on low-probability states that are still within the safe set. A natural alternative is to frame the problem as a robust optimization over a set of initial conditions that specify the initial state, dynamics and safe set, but whether this problem has a solution depends on the feasibility of the specified set, which is unknown a priori. We propose Feasibility-Guided Exploration (FGE), a method that simultaneously identifies a subset of feasible initial conditions under which a safe policy exists, and learns a policy to solve the reachability problem over this set of initial conditions. Empirical results demonstrate that FGE learns policies with over 50% more coverage than the best existing method for challenging initial conditions across tasks in the MuJoCo simulator and the Kinetix simulator with pixel observations.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Safe and constrained RL, one of 15 directions in Deep reinforcement learning methods. Builds on 43 of its 58 references that are in the map. First posted on arXiv 17 Feb 2026.

Builds on

43 of 58 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024Hamilton-Jacobi Reachability in Reinforcement Learning: A Survey12 Jul

Other references · 42

  1. 2017Proximal Policy Optimization Algorithmscited by 15,215
  2. 2015Variational Inference with Normalizing Flowscited by 2,170
  3. 2017Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real Worldcited by 1,498
  4. 2016Optimization Methods for Large-Scale Machine Learningcited by 1,150
  5. 2018Exploration by Random Network Distillationcited by 993
  6. 2018DeepMind Control Suitecited by 956
  7. 2021Deep Reinforcement Learning at the Edge of the Statistical Precipicecited by 807
  8. 2019Neural Spline Flowscited by 596
  9. 2017Robust Adversarial Reinforcement Learningcited by 451
  10. 2019Online Learning: A Modern Introduction Using Convex Optimizationcited by 417
  11. 2020Curriculum Learning for Reinforcement Learning Domains: A Framework and Surveycited by 373
  12. 2020Responsive Safety in Reinforcement Learning by PID Lagrangian Methodscited by 316
30 more
  1. 2020Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Designcited by 275
  2. 2015Contextual Markov Decision Processescited by 220
  3. 2020Projection-Based Constrained Policy Optimizationcited by 203
  4. 2020Prioritized Level Replaycited by 198
  5. 2022Evolving Curricula with Regret-Based Environment Designcited by 173
  6. 2020CRPO: A New Approach for Safe Reinforcement Learning with Convergence Guaranteecited by 134
  7. 2019IPO: Interior-point Policy Optimization under Constraintscited by 132
  8. 2021Replay-Guided Adversarial Environment Designcited by 111
  9. 2018Deep Counterfactual Regret Minimizationcited by 103
  10. 2020DeepReach: A Deep Learning Approach to High-Dimensional Reachabilitycited by 91
  11. 2020Automatic Curriculum Learning through Value Disagreementcited by 87
  12. 2022Saute RL: Almost Surely Safe Reinforcement Learning Using State Augmentationcited by 69
  13. 2021Safety and Liveness Guarantees through Reach-Avoid Reinforcement Learningcited by 49
  14. 2024Kinetix: Investigating the Training of General Agents through Open-Ended Physics-Based Control Taskscited by 45
  15. 2020Human-Level Performance in No-Press Diplomacy via Equilibrium Searchcited by 41
  16. 2023Iterative Reachability Estimation for Safe Reinforcement Learningcited by 37
  17. 2024No Regrets: Investigating and Improving Regret Approximations for Curriculum Discoverycited by 31
  18. 2022Lyapunov Density Models: Constraining Distribution Shift in Learning-Based Controlcited by 28
  19. 2022ISAACS: Iterative Soft Adversarial Actor-Critic for Safetycited by 28
  20. 2023Solving Stabilize-Avoid Optimal Control via Epigraph Form and Deep Reinforcement Learningcited by 24
  21. 2020Stochastic Regret Minimization in Extensive-Form Gamescited by 17
  22. 2023In-Distribution Barrier Functions: Self-Supervised Policy Filters that Avoid Out-of-Distribution Statescited by 16
  23. 2024Gameplay Filters: Robust Zero-Shot Safety through Adversarial Imaginationcited by 16
  24. 2024Solving Minimum-Cost Reach Avoid using Reinforcement Learningcited by 14
  25. 2024RACER: Epistemic Risk-Sensitive RL Enables Fast Driving with Fewer Crashescited by 10
  26. 2024To the Max: Reinventing Reward in Reinforcement Learningcited by 8
  27. 2025Solving Multi-Agent Safe Optimal Control with Distributed Epigraph Form MARLcited by 6
  28. 2023ROSARL: Reward-Only Safe Reinforcement Learningcited by 5
  29. 2022Feasible Adversarial Robust Reinforcement Learning for Underspecified Environmentscited by 4
  30. 2022Optimistic Gradient Descent Ascent in Zero-Sum and General-Sum Bilinear Gamescited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Beyond Hard Constraints: Budget-Conditioned Reachability For Safe Offline Reinforcement Learning
  2. 2026Robust Peak-cost Constrained Reinforcement Learning
  3. 2024Hyper: Hyperparameter Robust Efficient Exploration in Reinforcement Learning
  4. 2026Imagine to Ensure Safety in Hierarchical Reinforcement Learning
  5. 2026Parameter Exploration for RLVR via Variational Learning
  6. 2026Sampling-Based Safe Reinforcement Learning
  7. 2026SUN: Reaching for Novelty in Reinforcement Learning
  8. 2026Reinforcement Learning for Reachability: Guaranteeing Asymptotic Optimality