Orchard Agv Visual Navigation Research

With the development of the computer science and automatic control technology, AGV application has spread from industrial to service, medical treatment, military affairs, agriculture and forestry and other fields. In order to reduce arduous and heavy operation in orchard, the farmers urgently need automatic operation by machinery instead of heavy manual tasks. Orchard AGV is an intelligent vehicle, which can make autonomous traveling in orchard environment. It involves many theories and techniques in several fields. How to walk in orchard stably is a necessary basic research for AGV.As a key technology in AGV research, the goal of visual navigation is to accurately identify the path region and the non-path region in task space according to the information obtained from image, or to plan an optimized path to a given target according to the information of structured environment. In the orchard which is a complex task space with less structural characteristics, the visual navigation of the AGV will face more difficulties. This paper studies some problems on the AGV visual navigation in orchard, and presents three methods for object recognition of visual navigation in semi-structured environment and non-structured environment.The environment images are analyzed in RGB vector space to study the feasibility of lane detection for orchard AGV. The navigation path is successfully recognized from the background. Automatic thresholding and morphological image processing are used for preprocessing to obtain binary images and remove noise respectively. The guidance points are found from the average position of the navigation path by progressive scanning. When the path has the obvious curve structure, the best guidance lane is located using least squares curve fitting, otherwise detecting the straight line as the navigation base line. At lase the mathematical expression of the optical navigation base line is given.It is suitable to laying navigation marks to guide the orchard AGV, that's because the fruit plant in orchard is tall and plant spacing is long. This article presents a vision navigation method which is based on artificial navigation landmarks. With the artifichial landmarks the orchard AGV can obtain the the positon and orientation information accuracyly. When the artificial landmark is in the visual field, the accuracy position of the AGV will be located by the movement update and the perception update. In this method we use combination of the different words to represent different actions so as to order AGV make different moving.Line structural information is obviously in orchard especially when the ground surface is smooth. We locate points of trunk-and-earth intersection by machine vision and then to generate guidance line in orchard by using least square line fitting. Meanwhile the kinematics model of the AGV is established, the article established fixed-orientation and fixed-point moving fuzzy control system with the expert experience. Simulations through Matlab show that the fuzzy control system is effective.The experience shows that the algorithms are able to recognize the navigation paths robustly and detecete navigation successfully in the scenes. The three methods are mutual interdependence, if used in conjunction; they can promote the level of orchard automatic operation.