ADAPTIVE NEURO-FUZZY CONTROL DESIGN FOR TWO COOPERATING ROBOT MANIPULATORS HANDLING A FLEXIBLE BEAM

This paper presents an adaptive neuro-fuzzy logic control design by using singular perturbation approach for trajectory tracking of two cooperating planar rigid robots handling a flexible beam which does not require any vibration measurement of the beam. First, applying two-time scale control theory on the combined dynamics, a control scheme is elaborated which makes the position and orientation of the mass centre of the beam track a desired trajectory while suppressing the beam vibration. In the proposed controller, fast subsystem controller will damp out the vibration of the flexible beam by an optimal design and the other slow subsystem fuzzy controller dominates the trajectory tracking. The Neural Network is used to learn the rules weights of the fuzzy logic controller. To avoid requiring any beam vibration measurement for the fast control law, a linear time-variant observer is also designed. The simulation results show the Neuro-fuzzy techniques have been more successful than conventional control because they combine the learning capabilities of neural network with the transparent Knowledge representation of fuzzy controller.