Files
Abstract
Timely harvesting of quality cotton fiber is among the most pressing challenges in the cotton production industry. The current practice of mechanical harvesting after defoliation has increased acreage and production, but reduced efficiency of harvest. Farmers pick the cotton after at least 60% to 75% of the cotton bolls are open, at which point many of the earlier opening bolls have been exposed to weather more than 40 days waiting to be picked. Boll quality is compromised, and some bolls have already fallen to the ground, unharvestable. An additional problem is the availability and expense of a skilled labor force. The average age of a farmer in the U.S. is 60 years old, and that age has been increasing for decades. Thus, it is paramount to utilize the current nascent technologies in automation and robotics to develop revolutionary solutions to address these issues. This dissertation focuses on the development of the cotton harvesting robot to increase efficiency, save labor, and improve farming management.
A center-articulated and hydrostatic rover with an attached cartesian manipulator was designed and implemented. The robot integrated advanced sensing systems using encoders, a low-cost RTK-GNSS, a potentiometer, RGB stereo cameras, and IMUs to control navigation and picking manipulators. The robot also integrated three controllers to do advanced object detection, and control to harvest the bolls. Robot Operating System (ROS) was used to integrate and control the robotic system for cotton boll tracking, cotton boll location estimation, cotton rows detection, navigation, and harvesting. The sensor fusion algorithm Extended Kalman Filter (EKF) was utilized to perform autonomous localization and navigation of the robot. Cotton harvesting was achieved by using a ROS-independent finite state machine (SMACH), modified pure pursuit algorithm, and proportional–integral–derivative controller. The performance of the robot was evaluated and reported. Experimental results showed that the developed robot could precisely and efficiently navigate over the cotton rows and harvest the cotton bolls. Furthermore, the robotic design has shown that traditional vacuum harvesting can well be adopted in robotic systems to harvest the cotton bolls. The designed robot sets preliminary development success to improve cotton harvesting management.
A center-articulated and hydrostatic rover with an attached cartesian manipulator was designed and implemented. The robot integrated advanced sensing systems using encoders, a low-cost RTK-GNSS, a potentiometer, RGB stereo cameras, and IMUs to control navigation and picking manipulators. The robot also integrated three controllers to do advanced object detection, and control to harvest the bolls. Robot Operating System (ROS) was used to integrate and control the robotic system for cotton boll tracking, cotton boll location estimation, cotton rows detection, navigation, and harvesting. The sensor fusion algorithm Extended Kalman Filter (EKF) was utilized to perform autonomous localization and navigation of the robot. Cotton harvesting was achieved by using a ROS-independent finite state machine (SMACH), modified pure pursuit algorithm, and proportional–integral–derivative controller. The performance of the robot was evaluated and reported. Experimental results showed that the developed robot could precisely and efficiently navigate over the cotton rows and harvest the cotton bolls. Furthermore, the robotic design has shown that traditional vacuum harvesting can well be adopted in robotic systems to harvest the cotton bolls. The designed robot sets preliminary development success to improve cotton harvesting management.