Computer vision assisted semi-automatic virtual reality calibration

New developments of edge matching based semiautomatic virtual reality calibration are presented to provide more reliable and accurate telerobotic servicing. Three basic algorithms employed are 1) weighted average local edge detector, 2) weighted least squares for the edge-based camera calibration and object localization, and 3) robust matching to remove outliers. A key new 20-variable least-squares algorithm is developed that updates both the camera and the object models simultaneously for given two camera views of two mating objects. This simultaneous update reduces relative positioning errors considerably, and is essential to achieve high alignment precision. The developed system has been successfully tested for an ORU (Orbital Replacement Unit) insertion task within a +/-1/4 inch and +/-3 degrees alignment precision.