Robust Shielding for Safe Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Court, Edwin Hamel-De le, Badings, Thom, Abate, Alessandro, Belardinelli, Francesco, Fabiano, Francesco
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916068553916416
author Court, Edwin Hamel-De le
Badings, Thom
Abate, Alessandro
Belardinelli, Francesco
Fabiano, Francesco
author_facet Court, Edwin Hamel-De le
Badings, Thom
Abate, Alessandro
Belardinelli, Francesco
Fabiano, Francesco
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2606_00270
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Robust Shielding for Safe Reinforcement Learning
Court, Edwin Hamel-De le
Badings, Thom
Abate, Alessandro
Belardinelli, Francesco
Fabiano, Francesco
Artificial Intelligence
Machine Learning
Logic in Computer Science
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.
title Robust Shielding for Safe Reinforcement Learning
topic Artificial Intelligence
Machine Learning
Logic in Computer Science
url https://arxiv.org/abs/2606.00270