Cotton is one of the most economically important crops in India that is grown for its soft and fluffy fibers. The labor-intensive nature of cotton harvesting necessitates automating the task to speed up the process with reduced human labor. In countries like the USA and Australia, cotton is conventionally harvested using heavy and expensive harvesters once at the end of the growing seasons. The fiber in the early-opened bolls is left on the plants till the end of the season to be harvested, resulting in its exposure to weather and thus degraded quality. These issues can be addressed by using small robotic cotton harvesters, capable of harvesting cotton bolls at multiple times of the cotton plant life cycle. This paper presents the design and working principles of CottonHusker robot consisting of a 5-DOF (degrees of freedom) robotic arm with a custom gripper. This robotic arm is mounted on an autonomously navigating platform integrated with a deep learning-based perception module for fast and efficient automatic harvesting of cotton in controlled indoor environments. The single-shot multibox detector model was used following a comparative performance analysis of the state-of-the-art deep learning models. The accelerated SSD model along with a stereovision camera and an onboard computer comprised the perception module. The robot demonstrated promising cotton-picking performance, with 90% of picked material being cotton, showcasing the viability of the prototype in real-world field conditions. The robot’s ability to autonomously detect and harvest cotton in controlled laboratory conditions, while navigating around obstacles, marks a significant step toward the paradigm of smart agriculture.
CottonHusker: Deep Learning Enabled Cotton Picking Robot for Smart Agriculture
Springer Proc. Inf. Commun. Technol.
National Conference on CONTROL INSTRUMENTATION SYSTEM CONFERENCE ; 2023 ; Manipal, India October 06, 2023 - October 07, 2023
International Conference on Systems and Technologies for Smart Agriculture ; Kapitel : 51 ; 619-631
29.01.2025
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch