A Stereovision-based Crop Row Detection Method for Tractor-automated Guidance

Abstract

Steering agricultural machinery within rowed crop fields is a tedious task for producers. Automated guidance of the machinery will not only reduce operator fatigue but also increase both the productivity and safety of the operation. An essential aspect of automatic guidance is the ability to identify the pathway between the crop rows. This research was to develop an implementable row-detection algorithm for a stereovision-based agricultural machinery guidance system. The algorithm consists of functions for stereo-image processing, elevation map creation and navigation point determination. The method developed first reconstructed a three-dimensional crop elevation map from a stereovision image of crop rows and then searched for optimal navigation points from the map. The developed stereovision-based crop row detection system was tested in a soya bean field to follow both straight and curved soya bean rows at typical operating speeds. Field validation tests indicated that the stereovision-based guidance system could localise crop rows accurately and reliably in a weedy field with missing sections of soya beans. Based on crop row localisation information, an automated navigation system could guide an autonomous agricultural tractor following both straight and curved rows accurately at normal field operation speeds.