Tracker calibration using tetrahedral mesh and tricubic spline models of warp

Abstract

This paper presents a three-level tracker calibration system that greatly reduces errors in tracked position and orientation. The first level computes an error-minimizing rigid body transform that eliminates the need for precise alignment of a tracker base frame. The second corrects for field warp by interpolating correction values stored with vertices in a tetrahedrization of warped space. The third performs an alternative field warp calibration by interpolating corrections in the parameter space of a tricubic spline model of field warp. The system is evaluated for field warp calibration near a passive-haptic panel in both low-warp and high-warp environments. The spline method produces the most accurate results, reducing median position error by over 90% and median orientation error by over 80% when compared to the use of only a rigid body transform.