Uncertainty estimation for kinematic laser tracker measurements incorporating the control information of an industrial robot

Laser trackers are widely used in the field of industrial metrology to measure kinematic tasks such as tracking robot movements. In order to assess spatiotemporal deviations of a robots movement it is crucial to have a reliable uncertainty of the kinematic measurements. Common methods to evaluate the uncertainty in kinematic measurements include approximations specified by the manufactures, various analytical adjustment methods and Kalman filters. In this paper a new, real-time approach is proposed, which estimates the 4D-path (3D-position + time) uncertainty of an arbitrary path in space by incorporating the control information of an industrial robot. The proposed approach relies on Bayesian filtering and avoids the need for predefined motion models due to the incorporation of the robots control information into the analysis method. Furthermore, a common laser tracker model is augmented with kinematic properties in order to be utilized as a self-regulating measurement model, within the Bayesian filter. The new approach is demonstrated by using data obtained by tracking the tool centre point of a KR 5 ARC industrial robot with a synchronized Leica laser tracker AT901. It shows that the proposed approach is more appropriate to analysing kinematic processes, as it avoids all problems arising from predefined motion models in Bayesian filters for tracking. In comparison with the manufacturers approximations, the new approach takes account of kinematic behaviour, with an improved description of the real measurement process and a reduction in estimated variance. This approach is therefore well-suited to the analysis of robot movements.