Shields to Guarantee Probabilistic Safety in MDPs
Open Access
- 24 July 2026
- book chapter
- Published by Springer Nature
- p. 576-600
Abstract
Shielding is a prominent model-based technique to ensure safety of autonomous agents. Classical shielding aims to ensure that nothing bad ever happens and comes with strong guarantees about safety and maximal permissiveness. However, shielding systems for probabilistic safety, where something bad is allowed to happen with an acceptable probability, has proven to be more intricate. This paper presents a formal framework that conservatively extends classical shields to probabilistic safety. In this framework, we (i) demonstrate the impossibility of preserving the strong guarantees on safety and permissiveness, (ii) provide natural shields with weaker guarantees, and (iii) introduce offline and online shield constructions ensuring strong safety guarantees. The empirical evaluation highlights the practical advantages of the new shields, as well as their computational feasibility.Keywords
All Related Versions
This publication has 33 references indexed in Scilit:
- Prompt Runtime Enforcement2025
- Dynamic Shielding for Reinforcement Learning in Black-Box Environments2022
- Automata Learning Meets Shielding2022
- Online shielding for reinforcement learningInnovations in Systems and Software Engineering, 2022
- The probabilistic model checker StormInternational Journal on Software Tools for Technology Transfer, 2021
- Shield Synthesis for Reinforcement Learning2020
- Neural Simplex Architecture2020
- Shield synthesisFormal Methods in System Design, 2017
- Safety-Constrained Reinforcement Learning for MDPs2016
- Uppaal Stratego2015