Modelling and control of flexible manipulators

Structural flexibility is a characteristic of all robotic systems. Designing a high performance controller for a robot with a certain degree of link flexibility is not an easy task. This requires, not only a sound knowledge of the dynamics of the system to be controlled, but also a good characterization of the friction as well as an appropriate form of friction compensation. The presence of structural Vibrations in the system during and after high speed positioning makes the design of such a controller even more difficult. This thesis presents a method of modelling and controlling a single-link flexible manipulator for high speed precise end-point positioning. Several mathematicalmodels for a flexible robotic system are presented and described together with those for friction in the system. The parameters of the flexible system such as the natural frequencies, damping ratios and the parameters of friction, including Coulomb and Viscous friction coefficients, are identified offline, based on experimental data. Methods used for finding these parameters are described in detail. The objective of the control system is to move the flexible robot as fast as possible without producing vibrations at the end-point of the move. The position control of a single-link flexible robot is seen as a combination of two different problems namely trajectory planning and trajectory following. Two computational techniques, based on radial basis functions and the assumed-modes method, for planning the trajectories for a single-link flexible robot arm, are presented. These two techniques are compared with the aid of some numerical examples. In order to follow the derived trajectories, a combination of feedback and feedforward control along with a lead compensator is developed. The parameters of the compensator are found using an optimization process based on GeneticAlgorithms. Results of the experiments performed to verify the effectiveness of the proposed control strategy show that feedback control with feedforward compensation is more suitable than a simple feedback controller. Furthermore, these experiments also demonstrate the effectiveness of the trajectory planning process