Detection and three-dimensional localization by stereoscopic visual sensor and its application to a robot for picking asparagus

摘要

This paper describes a three-dimensional localization process used on an agricultural robot designed to pick white asparagus. The system uses two cameras (CCD and Newvicon).Thanks to diascopic lighting, the images can easily be binarized. The threshold of digitalization is determined automatically by the system. A statistical study of the different shapes of asparagus tips allowed us to determine certain discriminating parameters to detect the tips as they appear on the silhouette of the mound of earth.The localization is done stereometrically with two cameras. As the robot carrying the system moves, the images are altered and decision criteria modified. A study of the images from mobile objects produced by both tube and CCD cameras was carried out. A simulation of this phenomenon has been done to determine the modifications concerning object shapes, thresholding levels and decision parameters as a function of robot speed.