Component calibration and configuration planning in assembly automation with a parallel manipulator

Abstract

Parallel manipulator (PM) is widely used as posture alignment mechanism in heavy component machining and assembly process. To achieve precise kinematics backward transformation, the relative spatial relationship between the component and PM’s spherical joints need be calibrated, which suffers from the pain-point that the spherical joint center is inside the mechanical socket and its position can hardly be directly recovered. We refer this problem as component calibration. In this research, we present an alternate solution by formulating kinematic chain in form of AX = XB based on designedly established PM reference configuration and base coordinate system, which does not rely on the measurement of spherical joint center position. To improve the calibration accuracy, accuracy-influencing factors, PM’s dexterity distribution over workspace and properties of angle-axis space are jointly analyzed then 4 centrosymmetric angle-axis vector pairs on the boundary of maximum feasible angle-axis space at zero manipulator translation are set as configuration pairs to establish the kinematic chain. Configuration planning in angle-axis space facilitates the visualization and control of the critical factors in calibration process and tends to set these factors to optimum simultaneously. Experiments are carried out finally and the results show that the calibration accuracy could be notably improved with such method compared with no configuration planning, especially for translation accuracy. • A component calibration method in assembly automation with parallel manipulator. • Dexterity analysis and calibration configuration planning in angle-axis space. • A fast calibration implementation using fiducial tag and monocular vision.