Development of a multivariable fuzzy controller for site-specific herbicide applications in precision agriculture

A multivariable fuzzy controller for a multisection agricultural sprayer has been developed, making possible the site-specific management of crops, one of the fundamental goals in Precision Agriculture. The controller adjusts the herbicide dosage to be sprayed by opening/closing each independent section of the pulverization equipment, regulating the herbicide flow, and acting over the tractor velocity. The needed dosage at each point has been calculated using the information captured into two input maps: a) weed coverage map and b) state of growth map. The maps have been constructed by taking pictures in the field, at each sampling point of a predetermined grid spacing, and assessing the images in the laboratory. An artificial vision system has been developed to aid the user to compute, at each image, the percentages of pixels representing weed, crop and soil and obtain the maps. The controller has been implemented and tested using Matlab. The results indicate that, with the proposed approach, significant amounts of herbicide could be saved without losing efficacy on weed treatments. Copyright (C) 2007 CEA-IFAC.