A solution to pose ambiguity of visual markers using Moiré patterns

Abstract

Visual markers are useful tools assisting visual recognition of object pose in robotic applications. But they have two fundamental problems in orientation estimation. One is degradation of orientation accuracy in frontal observation. The other is “pose ambiguity” that the orientation cannot be determined uniquely. We previously developed a novel visual marker “LentiMark” which solves the former problem. This time we propose a solution to the latter problem by improving it. We utilize two new moiré patterns to get unique angle information and eliminate the pose ambiguity. The new marker enables stable and robust pose estimation without suffering from the pose ambiguities. The contribution of this study is that we solved the biggest two problems of visual markers, in the framework of small planar markers. We demonstrated the availability of our method through some comparative experiments.