The field of autonomous driving is moving towards a greater emphasis on safety and risk management. Researchers are exploring innovative approaches to ensure the safe operation of autonomous vehicles, including the development of risk-budgeted control frameworks, adaptive transition strategies, and game-theoretic risk-shaped reinforcement learning. These advances aim to balance safety and performance in complex traffic environments, where diverse agents interact and unexpected hazards frequently emerge. Notable papers in this area include:

• A risk-budgeted control framework that certifies safety for autonomous driving by switching between performance-based and conservative safety constraints.
• An adaptive transition strategy that dynamically adjusts control authority based on driver performance variations, reducing trajectory deviations and driver control effort.
• A game-theoretic risk-shaped reinforcement learning framework that incorporates a multi-level game-theoretic world model and an adaptive rollout horizon to ensure safe autonomous driving. These innovative approaches are expected to significantly improve the safety and efficiency of autonomous vehicles in various traffic scenarios.