Field Testing of the Autonomous Cotton Harvesting Rover in an Undefoliated Cotton Field

Thursday, January 9, 2020: 1:15 PM
401 (JW Marriott Austin Hotel)
Kadeghe Goodluck Fue , University of Georgia
Wesley M Porter , Assistant Professor, Department of Crop and Soil Sciences
Glen C. Rains , University of Georgia
Edward M. Barnes , Cotton Incorporated
Cotton bolls picked as they open would make it possible to use a team of smaller harvesting machines that use less energy, create less compaction, less machine downtime and can pick high quality bolls before deteriorated by the environment. But machines to do that job are not yet available. This study proposes the use of an autonomous robot to harvest the bolls as they open. We developed a cotton harvesting robot that is a center-articulated vehicle with four wheels and an x-y picking manipulator. The robot uses stereo camera to see rows, RTK-GPS to localize itself, fisheye camera for heading, and stereo camera to locate the cotton bolls. Using SMACH library which is the ROS-independent task-level architecture to build state machines for rapid implementation of the robot behavior. First, the GPS waypoints are obtained and then, the rover passes over the rows while picking the cotton bolls. The navigation is controlled by pure-pursuit together with PID controller. Two parallel programs organize the whole robot regarding which time to pick and which time to navigate. While navigating, the robot will be looking to find any bolls, and if it finds any, it will stop and pick. It will do this repetitive work until it finishes all the rows.