Workspace Trajectory Control of Flexible Robot Manipulators Using Neural Network and Visual Sensor feedback

This paper addresses the issue of workspace trajectory tracking control of flexible robot manipulators. A control strategy integrated with computed-torque-like control scheme, link vibration suppressing control scheme, and learning control scheme is proposed. Dynamics of a virtual rigid robot and vision feedback are used to construct the computed torque like controller. The learning controller is designed with a multilayer feedforward neural network to provide the control system with learning ability for improving performance of the control system. A vision system consisting of a CCD camera and video tracker is used for the measuring of end-effector position and link deflections. Simulations and experiments of end-effector trajectory tracking control are carried out using a 2-link flexible robot system as a testbed. The results confirm the effectiveness and usefulness of the proposed control strategy.