Option-based motion planning and ANFIS-based tracking control for wheeled robot in cluttered environment

Motion planning and trajectory tracking control are two fundamental problems needed to be solved when wheeled robots maneuver and operate autonomously in the cluttered environment. In this paper, two integrated and intelligent approaches are applied to solve these problems. Firstly, an option-based hierarchical reinforcement learning approach integrated with transfer learning is proposed to accomplish motion planning task in the cluttered environment, the transfer learning approach is employed to speed up the learning process. Then, the generated trajectory is tracked by an ANFIS-based controller, the parameters of inference system are updated online by gradient descent learning algorithm. The performance of using proposed intelligent approaches to control mobile robot in cluttered environment is validated in the simulation.