Algorithms for Objects' Shape Recognition and Measurement Before Gripping

Abstract

The paper proposes an algorithmic methodology to solve a necessary first step, namely visualization, recognition and measurement of objects, for gripping, in particular by means of anthropomorphic grippers for robots. For object shape recognition, Haar classifiers, and an Open CV library use are proposed. Based on measurement of the distance between the initial point and a second intermediate point, you can do calculations, aimed at object size, and distance remaining to the target. Thus, the algorithmic method for visualization, recognition, and measurement of objects for anthropomorphic gripping, is based on a number of existing software modules. They have been adapted for the intended purpose, and specially designed software modules were added.