Robust Shielding for Safe Reinforcement Learning
Quick Take
The paper presents a novel shielding framework for robust MDPs that ensures safety in reinforcement learning agents by satisfying LTL formulas under worst-case transition probabilities. This framework is sound and optimal, allowing for high-confidence safety guarantees while minimizing restrictions, even in unknown MDPs. Experiments demonstrate that as sample size increases, the expected return improves significantly.
Key Points
- Introduces a shielding framework for robust MDPs (RMDPs) with transition probability sets.
- Defines safety based on LTL formula satisfaction under worst-case scenarios.
- Proves the framework is sound and optimal for RMDPs, ensuring policy safety.
- Combines with sampling methods for learning transition probabilities with PAC guarantees.
- Experiments show safety guarantees improve as sample size increases.
Article Content
From source RSS / original summaryarXiv:2606. 00270v1 Announce Type: new Abstract: Shielding is an effective approach to formally guarantee the safety of reinforcement learning agents in Markov decision processes (MDPs). However, existing shielding techniques typically assume knowledge of the safety-relevant transition dynamics - a requirement that is seldom met in practice. To address this limitation, we introduce a novel shielding framework for robust MDPs (RMDPs), i. e. , MDPs with sets of transition probabilities.
We define safety as the satisfaction of a linear temporal logic (LTL) formula with a certain threshold probability under the worst-case transition probabilities of the RMDP. We prove that our shielding framework is both sound and optimal for the RMDP: every policy admissible by the shield is safe, and conversely, every safe RMDP policy is admissible by the shield.
We combine our approach with existing sampling methods for learning transition probabilities of MDPs with probably approximately correct (PAC) guarantees. This combination enables the construction of shields for MDPs that, with high confidence, guarantee safety while remaining minimally restrictive. Our experiments show that our shields for learned RMDPs guarantee safety in unknown MDPs while recovering strong expected return as the number of samples increases.
Reader Mode unavailable (could not extract clean content).
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →MindGames Arena Generalization Track: In2AI Solution with Delayed Per-Step Reward Attribution
The In2AI solution introduces delayed per-step reward attribution for training language model agents in multi-agent environments, achieving top performance on the MindGames Arena benchmark at NeurIPS 2025. An 8-billion-parameter model outperformed larger proprietary systems, including GPT-5, in competitive play, demonstrating enhanced stability and sample efficiency in reinforcement learning.
