Non-reproducible alignment and fitting algorithm effects on Laser Radar measurement

Ever-increasing introduction of new production technologies has significantly reduced manufacturing cycle time in recent years, especially joining technologies. For many industries, Zero Defect Manufacturing (ZDM) is considered as a key strategy to improve Right-First-Time (RFT) capability with a minimum waste of resources. There is a growing desire to move from off-line sample measurement to in-line data collection, which will only be possible with fast, accurate measurement technologies. Although metrology cycle times have improved with in-line measurement systems, their accuracy is not sufficient to meet the tight tolerance demands of typical high value manufacturing applications. A major obstacle to the uptake of new, non-contact measurement technologies is the difficultly in evaluating system capability in terms of repeatability, accuracy and calibration to recognized standards. This study considers these characteristics for a Laser Radar (LR) measurement system applied to an automotive door measurement task. To evaluate these factors, the authors consider: (1) the effect of tooling ball (TB) position and movement on part alignment and measurement feature results and (2) the feature-fitting algorithms applied to different sizes and orientations of hole. The results show that the statically-mounted LR is good at developing a repeatable coordinate system for the workpiece. Offsetting an individual TB had a statistically significantly effect on the repeatability of the measurement results. A number of feature-fitting algorithms were studied, with no algorithm providing a definitively superior result. Two data capture algorithm were considered; hatched and petal algorithm. The petal pattern algorithm is much faster, and was found to provide comparably repeatable results as the hatched pattern algorithm. These results give confidence that the LR system demonstrates good repeatability.