Can I read Embedding Safety into RL: A New Take on Trust Region Methods on EtoBox?

Embedding Safety into RL: A New Take on Trust Region Methods by Milosevic, Nikola; Müller, Johannes; Scherf, Nico is a scholarly article available to read on EtoBox.

What is Embedding Safety into RL: A New Take on Trust Region Methods about?

Reinforcement Learning (RL) agents can solve diverse tasks but often exhibit unsafe behavior. Constrained Markov Decision Processes (CMDPs) address this by enforcing safety constraints, yet existing methods either sacrifice reward maximization or allow unsafe training. We introduce Constrained Trust Region Policy Optimization (C-TRPO), which reshapes the policy space geometry to ensure trust regions contain only safe policies, guaranteeing constraint satisfaction throughout training. We analyze its theoretical properties and connections to TRPO, Natural Policy Gradient (NPG), and Constrained Policy Optimization (CPO). Experiments show that C-TRPO reduces constraint violations while maintaining competitive returns.

Author

Milosevic, Nikola; Müller, Johannes; Scherf, Nico

Published

2024

Language

EN