Jeeseop Kim

Safety Critical Control

Ensuring safety of robot locomotion in various conditions is essential to employ the system close to the human-centered environments. In the domain of motion planning and collision avoidance, various studies have been introduced on various platforms but not widely on legged robots. Furthermore, the safety-critical controllers typically leverage the distance from the unsafe set, resulting in robots maneuver safely only when in close proximity. To improve upon this, MPC approach with CBF or HJ reachability analysis have demonstrated the effectiveness with some computational burdens.

To date, my research includes developing the model-free safety-critical planner/controller subject to various environments including the coordination of the legged robots in various environments. Furthermore, this thread put the efforts to develop firm foundation of the notion of the safety to scale the application of safety-critical planner/controllers by leveraging tools from nonlinear control theory.

Research Aims

Research thread aims to establish systematic methods of achieving safety-critical robotic behaviors, including stable and robust legged locomotion in various environment setups, by leveraging tools from nonlinear control theory.