safe-rl

Introduces Boundary-Seeking Policy Gradient (BSPG), a first-order method for safe reinforcement learning that actively drives the policy toward the constraint boundary, with convergence guarantees and improved reward/boundary tracking on a Safety-Gymnasium task.

This paper introduces DROPJ, a human-centred method for safely training and deploying agent policies by learning a world model from real-world trajectories, then eliciting human preferences with justifications to train a reward model for model predictive control. Experiments show that using human-generated simulated trajectories and justifications improves safety and reduces computational cost.

SafeExplorer introduces an unbiased policy gradient estimator for reinforcement learning with recovery interventions, significantly reducing training-time falls on robot tasks while matching or exceeding standard PPO's final reward.

This paper proposes AutoSafe, a safety-aware policy architecture for safe online reinforcement learning that integrates structured safety monitoring and intervention directly into action generation, enabling smooth, risk-dependent transitions between performance and safety behaviors, demonstrated on benchmarks and a physical cart-pole system.

This paper proposes a safe transfer reinforcement learning framework for autonomous highway lane-changing, using adaptive teacher intervention and reward shaping to improve sample efficiency and safety. Experiments show over 52% improvement in safety and 5% improvement in efficiency over baselines.

TRIDENT is a novel multi-agent reinforcement learning framework that breaks the coupling between hybrid discrete-continuous actions, hard safety constraints, and physics-governed dynamics, achieving provably safe coordination with a convergence guarantee to a constrained Nash equilibrium and significant reductions in training-time violations.

This paper presents VLM-Safe-RL, a framework that integrates frozen vision-language models into constrained MDP Lagrangian updates to provide anticipatory cost signals for safe reinforcement learning in high-speed visual control tasks. The method outperforms standard constraint-aware baselines on Safety-Gymnasium FormulaOne L2 and generalizes to held-out environments.

Proposes LILAC+, a framework for safe continual reinforcement learning under nonstationarity that uses three adaptive safety mechanisms: context-based safety constraints, adaptation-speed constraints, and budget-to-state safety enforcement. Evaluations in simulated driving environments show reduced safety violations under distribution shift while maintaining competitive performance.

This paper proposes Action-Conditioned Risk Gating, a lightweight reinforcement learning method for risk-sensitive control under partial observability that uses a compact finite-history proxy state and an action-conditioned near-term risk predictor to balance safety and performance.

This paper presents a framework (CARE) that jointly learns control inputs and communication-efficient timing decisions under a pointwise Lyapunov safety shield, achieving higher inter-sample intervals than classical methods on inverted pendulum, cart-pole, and planar quadrotor systems.