Safe Reinforcement Learning via Recovery-based Shielding with Gaussian Process Dynamics Models

TL;DR

This paper presents a novel safe reinforcement learning framework that uses Gaussian process models for uncertainty quantification, enabling provably safe exploration and recovery in unknown nonlinear dynamical systems.

Contribution

It introduces a recovery-based shielding method combining a backup policy with GP-based safety prediction, ensuring safety guarantees while allowing efficient exploration.

Findings

Demonstrates strong safety and performance in continuous control tasks

Provides provable safety lower bounds for unknown systems

Enables sample-efficient learning with unrestricted exploration

Abstract

Reinforcement learning (RL) is a powerful framework for optimal decision-making and control but often lacks provable guarantees for safety-critical applications. In this paper, we introduce a novel recovery-based shielding framework that enables safe RL with a provable safety lower bound for unknown and non-linear continuous dynamical systems. The proposed approach integrates a backup policy (shield) with the RL agent, leveraging Gaussian process (GP) based uncertainty quantification to predict potential violations of safety constraints, dynamically recovering to safe trajectories only when necessary. Experience gathered by the 'shielded' agent is used to construct the GP models, with policy optimization via internal model-based sampling - enabling unrestricted exploration and sample efficient learning, without compromising safety. Empirically our approach demonstrates strong performance and strict safety-compliance on a suite of continuous control environments.