Safety-Critical Robotics and AI

Robotic and AI systems are unlocking transformative impact but also raising urgent safety questions. This course covers the mathematical foundations of safety-critical autonomy and introduces the key algorithmic paradigms (control barrier functions, Hamilton-Jacobi reachability, model-predictive shielding) under a unified safety filter framework. By dissecting the safety challenges in autonomous driving, robot learning, human interaction, and AI alignment, students learn the importance of quantifying uncertainty, minding the gap between models and reality, and ensuring safety - even under black swan events - without sacrificing performance.