An evaluation of artificial fiducial markers in underwater environments

This paper presents an evaluation of artificial fiducial markers in underwater environments. Due to the readily available code, the ARToolKit, AprilTags and ArUco marker systems are chosen for this purpose. The experiments are conducted in a tank equipped with a 3D gantry robot and a roll-tilt sensor head that allows to move the camera with millimeter precision. The markers are compared with each other at six turbidity levels in two lighting scenarios. The criteria are the smallest detectable marker size (in pixels), the maximum angle (camera-to-marker), the maximum distance and required processing time. The results show that the AprilTags library outperforms the other candidates for almost all tested criteria, but it also presents the slowest performance.