Current and Potential Robotic Applications to Improve Cotton Production

Friday, January 10, 2020: 8:00 AM
401 (JW Marriott Austin Hotel)
Edward M. Barnes , Cotton Incorporated
Gaylon Morgan , Texas A&M
Kater Hake , Cotton Incorporated
Jon Devine , Cotton Incorporated
Ryan W Kurtz , Cotton Incorporated
Terry W Griffin , Kansas State University
Gregory Ibendahl , Kansas State University
Ajay Sharda , Kansas State University
Glen C. Rains , University of Georgia
Kadeghe Goodluck Fue , University of Georgia
John L. Snider , University of Georgia
Matthew Aaron Bruce , University of Georgia
Alessandro Ermanis , University of Padova
Joe Mari J. Maja , Clemson University
Dennis Daly IV , Clemson University
Christina Chiu , Clemson University
Matthew Cutulle , Clemson University
Marlowe Edgar C. Burce , Clemson University
James A. Griffin , Texas A&M Cotton Extension
J. Alex Thomasson , Texas A&M University
Hussein Gharakhani , Texas A&M University
Emi Kimura , Texas A&M AgriLife Research and Extension Service
Brian G. Ayre , University of North Texas
Tyson B Raper , University of Tennessee
Sierra Young , North Carolina State University
Mathew G. Pelletier , USDA-ARS Cotton Production and Processing Research Unit
John D. Wanjura , USDA-ARS Cotton Production and Processing Research Unit
Greg A. Holt , USDA ARS CPPRU
Rapid advances in computer vision, deep learning, and autonomous equipment have resulted in commercial robots for autonomous weed control and point to a future where robotic systems will play a larger role in agricultural production.  Over the last two years, Cotton Incorporated has sponsored a small number of projects to look at the potential use of robots for cotton harvest and more recently, weed control. These projects have collectively resulted in an autonomous robot capable of traveling through the field using a combination of proximal sensors and GPS; a small tractor using machine vision that can autonomously harvesting cotton bolls under defoliated conditions; a draft economic model that incorporates regional climate data to compare the value of a robotic harvest system to once-over harvesters; and data from plots in four locations where cotton was hand-harvested twice per week to estimate the potential yield and quality benefits to provide estimates for the economic model.  Initial progress has also been made in evaluating new mechanism for removing seed cotton from the plant, developing an image library of weed species significant to cotton, exploration of genetic manipulations that could facilitate robotic harvest, and concepts around material handling in the field. In addition to providing an overview of recent progress, this paper will also explore other areas of cotton production that could benefit from the use of robotic systems.