Abstract
Point cloud registration is often required in processing measurement data and uncertainty is inevitably brought into this process. However, the effects of registration uncertainties have not been well studied in the literature, especially for high precision applications such as aircraft assembly. Take gap measurement in aircraft wing assembly as an example, using laser scanners and commercial software PolyWorks, this study investigates the registration uncertainty between part-level and assembly-level point cloud data, as well as related factors including system errors, sphere targets, and parameter settings. Results show that: (1) Use of sphere targets cannot improve registration accuracy in PolyWorks but can improve efficiency by reducing 41.94% of iteration times for small-scale point cloud data. (2) The registration process can bring errors and these errors can be affected by the parameter settings during data processing. (3) Systematic errors can be associated with both part-level and assembly-level measurements, these errors will be propagated during the registration process, and their effects on the gap measurement are dependent on the geometrical relationship of the two mating surfaces. A calibration method is proposed to mitigate the effects of systematic errors, and experiments were conducted to validate these methods. It is concluded that laser scanners and computational software can be used for high-precision assembly, and evaluating registration uncertainty is a crucial step for improving assembly accuracy.