Recently, there has been a growing interest in manipulating, transporting and positioning flexible payloads (for example, metal sheets) in various industrial applications, such as robotic assembly of automobile sheet-metal body parts. The system tackled in this thesis consists of a rigid robot manipulator and a piece of flexible sheet metal that is grasped at several points by the robot gripper. A major concern related to the performance of such a system is the vibration of the payload and its effect on the robot position. This vibration has to be eliminated before the payload can be precisely positioned for further processing on it. Thus, the main objective of this thesis is to develop an effective robot control strategy for payload vibration suppression that is practical in implementation and robust against model uncertainties.

Prior to developing a control scheme, it is useful to obtain an accurate and practical dynamics model of the system. To achieve this, a new method is presented in this thesis for modeling the entire robot-payload system. Using this dynamic model, the vibration controllability issue is examined in order to configure the robot-payload system properly so that all the critical vibration modes of the payload are controllable. Explicit criteria are provided to determine the configurations that result in uncontrollable vibration, and hence these situations can be avoided at the system design stage.

A feedforward/feedback control strategy is proposed in this thesis for payload vibration suppression. The feedforward loop consists of an input-preshaping technique and a computed-torque scheme. A new approach is developed for designing an input shaper. It utilizes a structure in which the modal forces of payload vibration are shaped in parallel. This approach relaxes the requirement of dynamic linearity in traditional shaper designs, and for the nonlinear robot-payload system, it improves the shaper's performance in vibration suppression. Given a precise dynamic model, the feedforward control can effectively achieve vibration-free movements of the payload. To enhance the control robustness against model uncertainties, a model-independent feedback loop is proposed, in which a robust vibration-suppression control is developed by using the reaction force of payload vibration at the robot gripper. The feedback control can accomplish further reduction of any residual vibration on the payload due to model uncertainties. This proposed control strategy is practical as it does not require direct measurement of payload vibration, and also has a simple structure with low on-line computational load.

Extensive numerical simulations are presented in this thesis to validate the developed dynamic models and to illustrate the effectiveness and robustness of the vibration-suppression control strategy. Rigorous stability and performance analyses of the control strategies are also conducted.