Prescribed Performance Control of Omnidirectional Mobile Robot Via Improved TD3 Algorithm

To meet high performance trajectory tracking control requirements of omnidirectional mobile robot, this study proposes a hybrid control architecture integrating prescribed performance control (PPC) and deep reinforcement learning (DRL). While traditional PPC can satisfy given performance constraints, its control performance cannot ensure the optimality. To this end, this paper introduces an improved Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm to optimize the controller parameters online of the proposed prescribed performance controller, thereby achieving a coordinated enhancement of stability and optimality.