Safety

Conservative and Adaptive Penalty for Model-Based Safe Reinforcement Learning
Conservative and Adaptive Penalty for Model-Based Safe Reinforcement Learning

February 2022

Conservative Offline Distributional Reinforcement Learning
Conservative Offline Distributional Reinforcement Learning

December 2021

Cautious adaptation for reinforcement learning in safety-critical settings
Cautious adaptation for reinforcement learning in safety-critical settings

How to train RL agents safely? We propose to pretrain a model-based agent in a mix of sandbox environments, then plan pessimistically when finetuning in the target environment.

July 2020