Autonomous Guidance of a Corn Harvester using Stereo Vision

Abstract

The growing demand for applications in Precision Agriculture has been proportional to the emergent interest in automatic guidance of agricultural machines. The difficult, and at the same time strenuous, task of driving a harvester during the prolonged working days of the harvesting season justifies the special attention paid to this technology. The development of Global Navigation Satellite Systems has meant an important push for autonomous navigation, especially in the case of off-road vehicles. However, experience has shown that automatic guidance inside the tight arrangement of crop rows is difficult to achieve unless local positioning sensors take part in the localization unit of auto-steered machines. This research develops a perception engine based on stereoscopic vision to detect the cut-uncut edge of corn during harvesting operations. The stereo camera was mounted on the head of an agricultural combine. The algorithm implemented was capable of finding the edge of corn, allowing the system to automatically guide the combine at regular harvesting speeds.