Specifying 3D Tracking System Accuracy One Manufacturer’s Views

Abstract

Manufacturers of 3D tracking systems use a wide variety of statistical measures, assessment protocols and measurement volumes when stating their systems’ accuracies. These factors typically differ ac- cording to the underlying technologies and the manufacturers’ personal preferences and experience, but because of competitive pressures, man- ufacturers tend to use protocols and statistical measures that empha- size their systems’ strengths and provide the best numerical values for comparisons. In addition, since 3D tracking systems generally have er- rors whose spatial distributions are nonuniform and which seldom follow known analytic distributions, the common practice of using a small num- ber of statistical measures to represent “typical” accuracies for these sys- tems is usually inadequate, and occasionally misleading. This can lead to a form of specmanship that can confuse potential users attempting to select the tracking systems best suited for their specific needs. We discuss some of the key accuracy factors often used to compare tracking systems, and we demonstrate some of the subtleties involved in accuracy specifications that potential customers should be aware of. The example systems cited are all manufactured by NDI.