Error Modeling and Accuracy of Parallel Industrial Robots

Abstract

Most industrial robots are open-chain mechanisms constructed of consecutive links connected by rotational or prismatic joints of one degree of freedom. These serial manipulators have large workspace, high dexterity and good maneuverability. However, due to their serial structure they exhibit low stiffness and poor positioning accuracy. As a result, their use in applications that require large loads (e.g. machining) and high accuracy, is limited. In the case of a parallel manipulator, the end-effector is attached to a moveable plate which is supported in-parallel by a number of actuated links. As a result, these parallel manipulators are anticipated to possess the following advantages, compared with serial manipulators: 1) high force/torque capacity since the load is distributed to several in-parallel actuators; 2) high structural rigidity; and 3) better accuracy due to less cumulative joint errors. A large number of publications dealing with the accuracy of the serial manipulators appeared in the past. These include topics on error modeling effects of manufacturing tolerance on pose accuracy and numerous calibration strategies. However, very few publications dealing with the same issue as related to parallel manipulators can be found. Since high accuracy is generally believed to be one of their advantages compared to that of serial manipulators, it is important to address this issue. The purpose of this research is to establish the kinematic and error models for evaluating the effects of manufacturing tolerances, installation errors and stiffness effect on the accuracy of a parallel robotic system. In order to evaluate the accuracy of parallel robotic system, it is necessary to develop a kinematic model which will accommodate the above errors. Based on this model, algorithms for forward, inverse kinematics and error modeling of the parallel robot are presented. These algorithms with a set of typical tolerances were used to compute the pose errors which include three translational and three angular errors.