Development and experimental verification of a mathematical model for robot force control design

Abstract

Describes the development and experimental verification of a mathematical model for a force control system consisting of a PUMA 560 manipulator, force sensor, and environment. The main contribution is an argument that explicit dynamic models of manipulation systems used for force control are of little relevance or use. Identification techniques are used to model the system as a black box, measuring only an input/output relationship. The explicit dynamics of the mechanical, hardware, and software elements of the manipulator are neglected; they are considered as parts of the measured input/output relationship. P+D loops are closed around each joint of the manipulator before identification to reduce the uncertainties, to overcome nonlinearities, and thus to linearize the system. Close correspondence between the experimental and the simulated Bode plots for frequencies below 30 Hz indicates that the model developed accurate enough to be used to analyze and design a force control strategy for hard-on-hard contact applications, e.g., robotic drilling.