The error propagation in robots.

Abstract

The accuracy of a robot manipulator has been receiving scrutinity since the widespread acceptance of robot manipulators. The relationship between two consecutive joint coordinate frames of a robot manipulator can be completely defined by five link parameters; one is the joint variable and the other four are the geometric parameters. The basis for the open-loop manipulator control is often the relationship between the Cartesian coordinates of the end-effector and the joint coordinates; therefore, the accuracy of the Cartesian position and orientation of the end-effector with regard to the real world depends on the errors of the five link parameters for each link. For design optimization and robot calibration, it is very important to develop a model for quantitative characterization and evaluation of the positioning and orientational errors of the end-effector. A static error propagation model is developed in order to describe the relationships between the six Cartesian errors and the five independent kinematic errors for each link. In this thesis, a general method for evaluating the end-effector errors produced by a mix of arbitrarily distributed errors is presented. Based on this method, any different combinations of biased and mixed error distributions can be dealt with directly to give a quantative error propagation analysis. Numerical results are presented for one, two and three degrees-of-freedom robot manipulators. Comparison of the results of the proposed model with other published model are presented and analyzed.