Crop signalling: A novel crop recognition technique for robotic weed control
Slaughter, David C.
Fennimore, Steven A.
Nguyen, Thuy T.
Vuong, Vivian L.
Smith, Richard F.
Siemens, Mark C.
AffiliationUniv Arizona, Dept Biosyst Engn
KeywordsControl and Systems Engineering
Agronomy and Crop Science
Animal Science and Zoology
MetadataShow full item record
PublisherACADEMIC PRESS INC ELSEVIER SCIENCE
CitationRaja, R., Slaughter, D. C., Fennimore, S. A., Nguyen, T. T., Vuong, V. L., Sinha, N., ... & Siemens, M. C. (2019). Crop signalling: A novel crop recognition technique for robotic weed control. Biosystems Engineering, 187, 278-291.
RightsCopyright © 2019 IAgrE. Published by Elsevier Ltd. All rights reserved.
Collection InformationThis item from the UA Faculty Publications collection is made available by the University of Arizona with support from the University of Arizona Libraries. If you have questions, please contact us at firstname.lastname@example.org.
AbstractWeed control is a significant cost for speciality crop producers, especially on organic farms. Agricultural operations are still largely dependent on hand weeding that is labour intensive and labour shortages and rising wages have led to a surge in food production costs. Thus, there is an inherent need to automate weed control and contain both labour costs and demands. Automatically distinguishing weeds from the crop plant is a complex problem since weeds come in a wide variety of colours, shapes, and sizes, and crop plant foliage is often overlapped with itself or occluded by the weeds. Current technology in commercial use, cannot reliably and effectively perform the differentiation task in such complex scenarios in real-time. As a solution to this problem, our team at the University of California, Davis has developed a novel concept called crop signalling, a technology to make crop plants machine readable and reliably distinguishable from weeds for automatic weed control. Four different techniques have been investigated and developed to make smart crop marking systems such as a) systemic markers, b) fluorescent proteins, c) plant labels and d) topical markers. Indoor experiments have been conducted for each method. Field experiments, using plant labels and the topical markers methods, have been successfully conducted for real-time weed control in tomato and lettuce. The results demonstrated that robots could automatically detect and distinguish 99.7% of the crop plants with no false positive errors in dense complex outdoor scenes with high weed densities. The crop/weed differentiation was thus effective, fast, reliable, and commercialisation of robotic weed control using the technique may be feasible.
Note24 month embargo; published online: 10 October 2019
VersionFinal accepted manuscript
SponsorsUSDA NIFA Speciality Crops Research Initiative [USDA-NIFA-SCRI-004530]; California Tomato Research Institute; California Leafy Greens Research Program