Abstract: This paper presents a constrained reinforcement learning (RL) framework for training a controller for nonlinear systems with guaranteed finite-time stability. The scheme is based on the deep ...