Comparison of serial and parallel robot repeatability based on different performance criteria

This paper is a contribution to the debate as to whether parallel robots are more precise than serial robots. To answer this question, the repeatability performances of three pairs of serial and parallel robots are compared with regard to several different criteria. The first pair is built from revolute actuated joints; the second and third pair from prismatic actuated joints. The serial and parallel robots have the same or almost the same workspace, the same actuators and the same covariance matrix. The repeatability modelling is based on the stochastic ellipsoid theory and enables us to compute the ISO9283 positional repeatability index in the whole workspace. Moreover, new geometrical criteria characterizing the ellipsoids are introduced: maximax and minimax semi-axes, area and eccentricity. We prove that the repeatability comparison results depend not only on the mechanism topology, the robot geometry, and the workspace location, but also on the choice of repeatability criterion. However, when robot design is optimised, the parallel robots under study herein have better repeatability performances.