Safe Reinforcement Learning via Probabilistic Logic Shields

2023/03/06 by Wen-Chi Yang, Giuseppe Marra, Yang, Wen-Chi +5 · 6 citations
Computer Science · Engineering · #Adversarial Robustness in Machine Learning #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Formal Methods in Verification #Machine Learning (cs.LG) #Safety Systems Engineering in Autonomy

paper · pdf · doi:10.48550/arxiv.2303.03226

openalex publication_date 2023/03/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Safe Reinforcement learning (Safe RL) aims at learning optimal policies while staying safe. A popular solution to Safe RL is shielding, which uses a logical safety specification to prevent an RL agent from taking unsafe actions. However, traditional shielding techniques are difficult to integrate with continuous, end-to-end deep RL methods. To this end, we introduce Probabilistic Logic Policy Gradient (PLPG). PLPG is a model-based Safe RL technique that uses probabilistic logic programming to model logical safety constraints as differentiable functions. Therefore, PLPG can be seamlessly applied to any policy gradient algorithm while still providing the same convergence guarantees. In our experiments, we show that PLPG learns safer and more rewarding policies compared to other state-of-the-art shielding techniques.

Cited by

• The Oversight Game: Learning to Cooperatively Balance an AI Agent's Safety and Autonomy
• ProSh: Probabilistic Shielding for Model-free Reinforcement Learning
• Exploring Structures of Inferential Mechanisms through Simplistic Digital Circuits
• Guardian: Decoupling Exploration from Safety in Reinforcement Learning
• Provably Optimal Reinforcement Learning under Safety Filtering
• LexiCon: a Benchmark for Planning under Temporal Constraints in Natural Language

Related

• Probabilistic Constraint for Safety-Critical Reinforcement Learning
• Cautious Reinforcement Learning with Logical Constraints
• Probabilistic Shielding for Safe Reinforcement Learning
• Think Smart, Act SMARL! Analyzing Probabilistic Logic Shields for Multi-Agent Reinforcement Learning
• ProSh: Probabilistic Shielding for Model-free Reinforcement Learning