Design and application of visual system in the Agaricus bisporus picking robot

Abstract

The Agaricus bisporus picking robot has not been commercially applied in the world. Based on the self-developed A. bisporus picking robot, the visual system design is carried out, and aiming at the specificity of the picking object, a measuring method of diameter and center point position based on monocular vision is proposed. Since the unequal heights of the A. bisporus affect the accuracy of visual recognition, the three-dimensional coordinates of the centre point of the A. bisporus are measured by the horizontal movement of the camera and the ellipse-fitting algorithm is proposed to improve the accuracy of position. Moreover, the depth information is used to compensate the error of the diameter measurement of A. bisporus. Through the picking experiment at the A. bisporus planting base, the results prove that the recognition success rate of the visual system is up to 90%.