Skip to main content
ARS Home » Southeast Area » Mississippi State, Mississippi » Crop Science Research Laboratory » Genetics and Sustainable Agriculture Research » Research » Publications at this Location » Publication #382229

Research Project: Closing the Yield Gap of Cotton, Corn, and Soybean in the Humid Southeast with More Sustainable Cropping Systems

Location: Genetics and Sustainable Agriculture Research

Title: Developement and optimization of a deep-learning-based egg collecting robot

item LI, GUOMING - Mississippi State University
item CHESSER JR, GRAY - Mississippi State University
item Huang, Yanbo
item ZHAO, YANG - University Of Tennessee
item Purswell, Joseph - Jody

Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/31/2021
Publication Date: 10/20/2021
Citation: Li, G., Chesser Jr, G.D., Huang, Y., Zhao, Y., Purswell, J.L. 2021. Developement and optimization of a deep-learning-based egg collecting robot. Transactions of the ASABE. 64(5):1659-1669.

Interpretive Summary: Robotics has been developed for automation of agricultural operations. This paper is a collaborative investigation of scientists in Mississippi State University, USDA ARS Genetics and Sustainable Agriculture Research Unit, University of Tennessee and USDA ARS Poultry Research Unit for integrating deep learning schemes to develop and optimize a robot arm to automatically identify and collect floor eggs in cage-free hen housing systems. The results indicated that the robot could accurately identify the eggs and, based on the egg identification, it could accomplish over 90% success rates in picking various eggs. The research of this paper provides an example of developing a robot integrated with deep learning from artificial intelligence for poultry farm management, and at the same time the information can be transferred to and shared with other agricultural research, for example, robotics for precision crop farming through deep learning-based detection and operation.

Technical Abstract: Manual collection of floor eggs in cage-free (CF) hen housing systems is time-consuming and laborious. The objectives of this research were to 1) develop a robot arm to automatically collect floor eggs and 2) optimize the performance of recognizing and picking eggs via the robot. The robot consisted of a deep-learning-based egg detector, a robot arm, a two-finger gripper, and a hand-mounted camera. The deep learning model, You Only Look Once (YOLO) V3, was embedded into the vision system to detect and locate eggs on a simulated litter floor in real time. Image processing algorithms (e.g., cropping, scaling, erosion, etc.) were implemented for the detection and provided the robot with centroid coordinates, orientation, and axis length of detected eggs, so that the gripper can be manipulated with fitted angles and appropriate openings to grasp detected eggs. For optimization, the YOLO V3 was retrained with the customized dataset of floor eggs and achieved over 93% performance on detecting and locating eggs; the kernel sizes of 65×65 pixels for erosion and dilation in image processing assisted in extracting geometry features of eggs with the least remaining noises; and, among the four testing cases , the soft grouting sponge attached to the gripper had the highest success rates for egg picking. After optimization, the robot accomplished 92-94% success rates in picking white and brown eggs. In sum, the developed egg collecting robot can be relied on for picking floor eggs to assist precision management in CF hen housing systems.