Depth estimation of objects with known geometric model for IBVS using an eye-in-hand camera

Abstract

One of the most crucial and challenging issues in Image-Based Visual Servoing (IBVS) is the derivation of accurate depth information of feature points, which is indispensable in the calculation of the image Jacobian. For an object without a known geometric model, it is difficult to retrieve the depth information of feature points using a single camera. However, in many application scenarios such as industrial manufacturing processes, geometric models of artificial objects such as polyhedrons are known in advance. As a result, this paper aims at developing an estimation algorithm that can be used to retrieve the depth information of an object with a known geometric model. Several computer simulations and real experiments have been conducted to verify the effectiveness of the proposed depth estimation algorithm. In addition, a 6-DOF industrial robot is used to perform an eye-in-hand IBVS task with the aid of the proposed depth estimation algorithm.