JMLR: OmniSafe is an infrastructural framework for accelerating SafeRL research.

Feasibility Consistent Representation Learning for Safe Reinforcement Learning (ICML 2024). Current SOTA model-free safe RL algorithm on safety-gym

Safe Reinforce Learning -> Constraint algorithms to train agents in Safety Gym, paper notes on research papers regarding RL with constraints + optimizer + neural networks, PyTorch implementation on policy gradient algorithms