PREDICTION OF THE DYNAMIC-RESPONSE OF FLEXIBLE MANIPULATORS FROM A MODAL DATABASE

Presented in this work is a modal-data-based and experimentally-orient ated method to predict the dynamic response of a time-varying and non- linear system, consisting of a robot manipulator with elastic members. The method utilizes modal models which can be reliably obtained throu gh well-known modal testing techniques. For such a system where modal characteristics vary with robot position or time, modal testing usuall y yields modal data in a limited number of positions. In order to empl oy these data, which may be coarsely placed in a robot's workspace, an interpolation technique is used to obtain continuous modal variations . These interpolated modal data along with the mass properties are uti lized to form a modal summation solution for prediction of the time hi story of deformable motion. This approach is more efficient since the modal database identified for a robot can be repeatedly used for a var iety of different tasks. Because modal data can usually be experimenta lly obtained from modal testing with a great degree of reliability, th e prediction on this basis should lead to more accurate time response information. Case studies are conducted for robot trajectory planning of a prismatic-and-revolute manipulator, with primary consideration of elastodynamically-induced positioning inaccuracy. (C) 1997 Elsevier S cience Ltd.