Collision Regulation

Collision regulation research focuses on developing robust and reliable methods for autonomous systems to safely navigate and avoid collisions, particularly in complex or uncertain environments, adhering to regulations like the COLREGs. Current approaches leverage diverse techniques, including neural networks (e.g., Signed Distance Fields within Quadratic Programs for reactive control), probabilistic methods for interpreting encounter situations under uncertainty, and deep reinforcement learning for risk-based decision-making and COLREGs compliance. This work is crucial for enabling the safe deployment of autonomous vehicles, particularly in maritime settings, and contributes significantly to both theoretical advancements in control and AI and practical applications in autonomous navigation.