Vision based position control of a robot manipulator using wavelet based feature extraction

In this paper, we present a novel technique for wavelet based corner detection as image feature points in a planar object for aligning a robot gripper with the object. Wavelets are suitable for the detection of sharp changes and locations of high curvature points to analyze local features in signals. In order to extract feature points for vision based position control, the polygonal approximation method is employed, which is composed of local maxima extracted from the wavelet transform of the object contour signals. These image features are used as final targets to guide the eye-in-hand robotic system to attain the desired orientation and height from which the gripping action can be executed. Experiment results for FARA robot are presented to show the applicability of this approach.