2D 1/2 Visual Servoing

Abstract

In this paper, the problem of estimating the partial camera displacement from two images of a static object is studied. This scaled displacement is used to design a new vision-based control scheme, called 2D 1/2 visual servoing. It must be emphasized that motion and structure can be estimated without the a priori knowledge of a 3D model of the object, which increases the versatility and the application area of visual servoing. Since vision-based robotics tasks generally necessitate to be performed at video rate, we focus only on linear algorithms. The classical approach to linearly estimate the motion parameters is based on the computation of the essential matrix. In this paper, we propose another approach, based on the estimation of an homography matrix related to a chosen plane of an unknown object. Simulations and experiments on a real scene show that this method gives a more robust reconstruction of the motion parameters, especially in the singular cases. We then focus on the 2D ...