Proposal of international voluntary activities on establishing benchmark test schemes for AR/MR geometric registration and tracking methods

This is a proposal to the ISMAR community from Japanese AR/MR researchers for the future progress of AR/MR technology. We hope to expand our activities over the ISMAR community and call for international participants who would take part in a number of voluntary works. At the same time, this paper presents a current view of the outcomes of these activities may have be in due course. We will focus here on the various tracking methods, one of the most active themes at the annual ISMAR symposiums. The main goal of our activities is to build a framework to comprehensively evaluate a variety of existing and future tracking methods. Strictly speaking, our targets should include all the geometric registration methods that merge the real and virtual world seamlessly. They can also be termed as real-time 3D matchmove. After initial registration (calibration) of two spaces is achieved, either object tracking or camera tracking is required when the subject or camera moves. In addition to methods suited to static registration, there exist also methods that focus solely on improving the performance of tracking, without calibration. In this paper, when we use the terms "registration and tracking" or simply "tracking" we will be referring to the general definition of tracking as given above.