Shape features for image-based servo-control using image moments

Abstract

The objective of the visual servoing (VS) is to control a dynamic system using the data extracted by a vision sensor. In Image Based Visual Servoing (IBVS) scheme control values are computed on the basis of image features directly. It is used in many applications such as automated machining, remote hazardous material handling, remote surgery and many more. In IBVS, singularities in the interaction matrix and local minima in the trajectories can severely affect the visual servoing task. There are two main challenges on IBVS referring to recent research. The first most important issue is the design and selection of visual features to avoid the singularity issue, and second is the development of a control scheme by selected visual features such that the desired characteristics are obtained during the visual servoing. Moments have been useful to select a minimal number of features to control the whole DOFs of a camera. Our objective is to obtain an interaction matrix with minimal set of features using image moments which change slowly around the desired position of the camera. In this paper we propose use of shape feature circularity, which is also invariant to translation, rotation and scale in IBVS control scheme. Initially analytic form of interaction matrix related to the selected visual features for planar object is derived for 6 combinations of features using moments to control the robot with 6 DOF and performance is evaluated through simulation results via machine vision toolbox of Matlab.