Headland turning of autonomous robot in corn field

Abstract

Headland turning operation has been difficulties in field application for autonomous agricultural vehicles based on machine vision. A method using variable field of view of camera to solve deficiencies in blind spots or limited image information with a settled field of view, which is controlled by two DC motors to rotate in vertical and horizontal planes, is proposed to detect crops near the end of crop rows, guide an agricultural robot and implement headland turning operation in corn field. The lateral field of view of camera was chosen firstly to detect crops near the end of corn row and guide the robot according to guidance information from crop images by using the image processing technology based on mathematically morphological operations. Headland turning was mainly composed of two 90% turning and one backing control. Three repetition tests were conducted to perform the headland turning operation in corn field. A maximum error of 17.4mm when using the lateral view and good headland turning operation were observed. It was successful for variable field of view to implement headland turning operation in corn field. © 2011 Springer-Verlag.