Methods to improve the dimensional measurement accuracy of a motion tracking system

摘要

Photogrammetry-based motion tracking systems are widely used in entertainment, medicine, sports, and other industries for low accuracy positioning applications. Camera-based systems track cooperative markers on objects to obtain their real-time positions and poses. There are few detailed studies in the literature on the metrological performance in three-dimensional (3D) space for such systems since they were primarily used in the entertainment industry and their off-the-shelf accuracy was not a key concern. However, applications for these systems have become more widespread due to their low cost and high repeatability (on the order of a few micrometers), and there is promise for applicability in metrology grade applications after suitable modifications. In this context, we investigate the error sources of a motion tracking system by error mapping in 3D space using a coordinate measuring machine (CMM) with two orders of magnitude lower positioning error. With the understanding gained from this study, we propose three different methods to improve the accuracy of motion tracking systems: interpolation based on an experimentally determined error map, error correction based on scale bar measurements, and weighted averaging using two motion tracking systems. Verification experiments are presented for each method. Our study is focused on the baseline errors of the motion tracking system, i.e., the system is used in the static mode where the coordinates are recorded when the probe (composed of markers) is stationary, not moving.