Detecting System of Contact Network Wrist Arm Bolts with 3D Template Matching and Visual Servoing Control

In this paper, a Position-Based Visual Servoing (PBVS) method based on 3-dimensional (3D) template matching technique is presented. Considering that traditional PBVS techniques are typically used to track planar (2D) objects, the tracking performance is worse for stereo (3D). The algorithm presented in this paper is a fusion and 3D pose estimation based on visual cues. The parameters of this transformation are estimated using a nonlinear minimization of a unique criterion that integrates information about the edges of the tracked object to efficiently estimate the pose information of a stereo object. The method has been validated in visual servoing experiments for bolt tracking of contact network components. The results show that the method has good performance in visual servoing tracking of 3D objects.