A 3D Vision Tracking Method for Mechanism Validation

Abstract

3D tracking of mechanisms in a non-structured environment is a very challenging task. It is important for the development of robotic applications, since key points can be tracked and used to control or to validate the system. In this paper, a 3D object-tracking method is presented. The proposed method allows to track a marker or a specific point also in poorly-lighted environments through images captured by a camera in combination with a depth sensor, thus obtaining the 3D Point Cloud of the entity as result. In addition, the velocity and the trajectory of the marker can be obtained. Two different acquisitions are reported as example, one for a cable-driven rehabilitation device and one for a lower-mobility parallel mechanism.