Committee: 

Dr. Chatterjee, Advisor  

Dr. Mukhopadhyay, Chair

Dr. Egerstedt

Abstract: The objective of this research is to investigate safe learning-based control for safety-critical robots while considering unmodeled system dynamics. Uncertainty in system dynamics can arise due to external disturbances or internal system failures. Uncertainty modeling is done with the help of posterior surrogates from Gaussian Processes (GPs). The safety framework leverages properties of Control Barrier Functions (CBFs) to ensure forward invariance of the system inside a safe set.