Neural-Network-Based Control of Wheeled Mobile Manipulators With Unknown Kinematic Models

As we all known, tracking control is always a basic and important issue in robot industry. In the past, the method based on the pseudo-inverse of Jacobian matrix is a conventional solution which has two major limitations: one is the position error accumulation phenomenon; the other is the requirement for the model information of robots. Therefore, in this paper, we propose a neural-network-based control (NNBC) scheme for wheeled mobile manipulators. The NNBC method overcomes the two limitations and is proved feasible to control the track of stationary robot manipulators. However, in robot industry, mobile manipulators controlling is an unavoidable topic. It is necessary to confirm the NNBC method is workable for mobile manipulators. Thus in this paper, we are going to discuss how to control the track of wheeled mobile manipulators using the NNBC method. In details, we will control the joints of the stationary manipulator and the wheels of the mobile platform to ensure the end-effector make a correct track without knowing the information of the parameters and structure of wheeled mobile manipulators. By integrating the stationary manipulator and the mobile platform into a system, the wheels and the manipulator can cooperate correctly to complete the task of the end-effector. Finally, some simulations will be performed to prove that the NNBC method is workable for wheeled mobile manipulators.