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-gymnasium

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

Adaptive risk-aware reinforcement learning framework for time-constrained navigation tasks. This MSc thesis studies how an RL agent can adapt its risk attitude when mission time changes, using a hierarchical controller that switches from a risk-aware CVaR-constrained policy to a risk-seeking policy when time pressure makes switching benificial.

A safety-critical Reinforcement Learning environment for Industrial Power Grid management, built with FastAPI and OpenEnv.