Research on Object Recognition of Intelligent Robot Base on Binocular Vision

Abstract

For intelligent robots need reliable object recognition and precise orientation in complex environments, this paper presents a method that use binocular vision to object recognition. In this paper, use least-squares fitting method to accurately determine coordinates of one matching point and four boundary points. calculate the three-dimension coordinate of the mathcing point via binocular vision theory, compute the three-dimension information include the size and height of the object via projection theory and restrictive relation between four boundary points and depth of mathcing point, which improve the reliability of object recognition and precise of orientation. The results of experiment show that this method can get reliable object recognition and precise orientation, meet the needs of robot path planning and grab objects with gripper.