CNU Professor Son Hyoung-il’s research team developed a human-centered robot that harvests fruits and vegetables by imitating human harvesting behavior. A human-centered robot refers to a robot that can assist humans in various environments.
 
The research team succeeded in developing a human-centered fruit and vegetable harvesting robot by technologically implementing various actions of workers harvesting greenhouse fruits and vegetables such as tomatoes and cucumbers.
 
When harvesting fruits and vegetables, human harvesters recognize the crops to be harvested, determines the harvesting order, and harvest the crops one by one. At this time, even if the harvested crops are shaking, the crops are held with one hand and the tops of the crops are cut off with the other hand using a tool, while still being visually aware of the shaking of the crops.

The fruit-and-vegetable harvesting robot developed by the research team replaces the behaviors displayed by human workers when harvesting by perception and conducts harvesting ordering, visual servoing, and gripping to assist harvesting. 
 
The research team implemented the “recognition” part of this robot using RGB-D camera-based deep learning so that it is not greatly affected by changes in the surrounding environment, such as light level. The Harvest Sequence Optimization section received the recognition results and optimized them into a 3D harvest path based on a genetic algorithm. Even if the crop to be harvested is shaken by various factors, real-time tracking control is performed by applying an image stabilization algorithm and a vector extraction algorithm of six motion elements of the crop top. Finally, the harvested crops were gripped and cut with a soft robot-based gripper consisting of a gripping module and a cutting module, and the same process was performed for the next harvesting sequence.
 
This study demonstrated a very high harvest success rate on a smart farm growing a large number of tomatoes and cucumbers. The research team is conducting follow-up research on harvesting robots that use both arms like human workers to improve harvesting speed and the harvesting success rate.
 
This research was participated in by Professor Son of the Department of Convergence Biosystems and Mechanical Engineering, as well as doctoral student Park Yong-hyeon, doctoral student Seol Jae-hwi, master’s students Park Jeong-hyeon and Jo Yoo-seong, and Kim Chang-jo from undergraduate-master’s curriculum.

The research results were published in Computers and Electronics in Agriculture (IF 8.3, JCR top 0.9%) and Precision Agriculture (IF 6.2, JCR top 7.8%), the most authoritative international academic journals in the field of convergence agriculture. This research was conducted with support from the Smart Farm Multi-Ministry Package Innovation Technology Development Project and BK21 Stage 4 IT-Bio Convergence System Agricultural Education Research Group.