Provably Safe Reinforcement Learning: Conceptual Analysis, Survey, and Benchmarking

2022-05-13 · Hanna Krasowski, Jakob Thumm, Marlon Müller, Lukas Schäfer, Xiao Wang, Matthias Althoff

Ensuring the safety of reinforcement learning (RL) algorithms is crucial to unlock their potential for many real-world tasks. However, vanilla RL and most safe RL approaches do not guarantee safety. In recent years, several methods have been proposed to provide hard safety guarantees for RL, which is essential for applications where unsafe actions could have disastrous consequences. Nevertheless, there is no comprehensive comparison of these provably safe RL methods. Therefore, we introduce a categorization of existing provably safe RL methods, present the conceptual foundations for both continuous and discrete action spaces, and empirically benchmark existing methods. We categorize the methods based on how they adapt the action: action replacement, action projection, and action masking. Our experiments on an inverted pendulum and a quadrotor stabilization task indicate that action replacement is the best-performing approach for these applications despite its comparatively simple realization. Furthermore, adding a reward penalty, every time the safety verification is engaged, improved training performance in our experiments. Finally, we provide practical guidance on selecting provably safe RL approaches depending on the safety specification, RL algorithm, and type of action space.

📄 PDF Abstract BibTeX arXiv:2205.06750

Code (0)

등록된 구현이 없습니다.

Tasks

Benchmarking reinforcement-learning Reinforcement Learning Reinforcement Learning (RL)Safe Reinforcement LearningSurvey

Is Q-Learning Provably Efficient? An Extended Analysis

2020-09-22 · Kushagra Rastogi, Jonathan Lee, Fabrice Harel-Canada, Aditya Joglekar

This work extends the analysis of the theoretical results presented within the paper Is Q-Learning Provably Efficient? by Jin et al. We include a survey of related research to contextualize the need for strengthening the…

Q-Learning reinforcement-learning Reinforcement Learning (RL)

Provably Safe Reinforcement Learning from Analytic Gradients

2025-06-02 · Tim Walter, Hannah Markgraf, Jonathan Külz, Matthias Althoff

Deploying autonomous robots in safety-critical applications requires safety guarantees. Provably safe reinforcement learning is an active field of research which aims to provide such guarantees using safeguards. These sa…

reinforcement-learning Reinforcement Learning Safe Reinforcement Learning

Safe Continual Reinforcement Learning Methods for Nonstationary Environments. Towards a Survey of the State of the Art

2026-01-08 · Timofey Tomashevskiy arxiv

This work provides a state-of-the-art survey of continual safe online reinforcement learning (COSRL) methods. We discuss theoretical aspects, challenges, and open questions in building continual online safe reinforcement…

Reinforcement Learning Continual Learning

Provably Efficient Iterated CVaR Reinforcement Learning with Function Approximation and Human Feedback

2023-07-06 · Yu Chen, Yihan Du, Pihe Hu, Siwei Wang 외

Risk-sensitive reinforcement learning (RL) aims to optimize policies that balance the expected reward and risk. In this paper, we present a novel risk-sensitive RL framework that employs an Iterated Conditional Value-at-…

Decision Making LEMMA reinforcement-learning Reinforcement Learning (RL)

Provably Safe Reinforcement Learning via Action Projection using Reachability Analysis and Polynomial Zonotopes

2022-10-19 · Niklas Kochdumper, Hanna Krasowski, Xiao Wang, Stanley Bak 외

While reinforcement learning produces very promising results for many applications, its main disadvantage is the lack of safety guarantees, which prevents its use in safety-critical systems. In this work, we address this…

reinforcement-learning Reinforcement Learning Reinforcement Learning (RL)Safe Reinforcement Learning