Kai S. Yun

Kai Yun is a graduate student in mechanical engineering whose work addresses fundamental safety challenges for autonomous robots. His work bridges the critical gap between rigorous theoretical guarantees and physical real-world deployment. By integrating control theory, deep learning, and large foundation models, he develops scalable control architectures for complex, high-dimensional systems, including quadruped and humanoid robots. He developed a framework for provably safe and scalable reinforcement learning, ensuring strict constraint satisfaction in high-dimensional systems. He also tackles uncertainties in system dynamics and perception modules to maintain reliability under out-of-distribution scenarios. MATLAB’s optimization infrastructure and semidefinite programming solvers are instrumental to his work on formal synthesis and verification of these control architectures. As a MathWorks Fellow, Kai will scale formal synthesis of safe controllers to highly dynamic humanoid systems and develop neural-control integration frameworks. Overall, his research lays the foundations for translating theoretical safety guarantees into practical, trustworthy autonomous robot deployment across diverse industries.