Input Laws for Dynamic Control: Theory, Experiments and Robustness

In the automation field is of significant economical value to accomplish accurate and fast positioning of handling devices. Examples are given by some robot manipulators, disk drive heads, pointing systems in space, etc. Many of them are made of flexible and light structures to manoeuvre as quickly as possible. This paper deals with the kinematic and dynamic control in the point-to-point motion of a deformable system. Prior to this work there are, by the same authors, a number of studies concerning with the problem of reducing the residual (final) vibrations in point-to-point movements. This article presents an important extension of the former works: not only the rest condition is considered as a desirable final position, but any (admissible) kinematic condition may be required. The adopted control technique is open–loop and it is concerned with the pre–shaping of the input law in the form of a limited number of a piece-wise acceleration/force/torque steps. In fact the so called “constant acceleration laws” proved themselves to be suitable for this purpose and showed many advantages over other pre–shaping techniques. This work also provides an extension of the previously mentioned approach in order to add good insensitivity to errors in modelling parameters without involving complex and expensive computations, giving in this way a deeper insight thought robustness. An experimental set with results is also described.