Experimental-numerical approach to the estimation of uncertainty in robot pose measurement

Abstract

Accurate pose measurements are crucial for the reliability of the kinematic calibration and performance evaluation of a robotic manipulator. Spherical coordinate measurement systems such as laser trackers and total stations are commonly used to collect sets of 3D points from which position and orientation of a moving coordinate frame attached to the end-effector in relation to a reference coordinate frame are calculated. Estimating the uncertainty of this specific measurement task is necessary to provide traceability to these measurements. In this paper we propose combing measurements of calibrated artefacts with Monte Carlo simulations to estimate this task-specific measurement uncertainty. The application of this method has been demonstrated for the performance evaluation of a Stewart Platform using a robotic total station. Expanded uncertainties between 0.40 mm and 0.63 mm have been estimated for position. Expanded uncertainty was 0.10° for all orientation results.