Modeling Activities During Manual Fruit Harvesting: A Comparison of Processes In the U.S. and Greece

Work method analyses are useful for improving production efficiency, operations management, and identifying differences in economic and environmental performance. This paper describes analyses and modeling of procedures during hand harvest of sweet cherries, considered as an industrial procedure, using the state diagrams of Unified Modeling Language (UML). First, the harvest process is split into two discrete parts: α) picking, and β) loading (collecting full fruit bins from the orchard). Second, the differences between the harvesting procedure in USA and Greece are highlighted and then models are presented. The discrete part (α) is executed in nearly the same way for both countries – only the capacity of the receptacles differs (20 kg in Greece vs. 200 kg in the U.S.). The procedure for loading differs however. In Greece, all bins along adjacent tree rows are removed from the grove and loaded manually onto a platform driven between tree rows; in the U.S. small tractors push a hydraulic bin trailer that retrieves up to 4 bins. UML state diagrams were used to model different states that an object of a class (a picker or a machine) could be in and explain how each object transitions from state to state. Each diagram represents objects of a single class and tracks the different states of its objects through the system dynamically. Our main emphasis is on the construction of rigorous models that provide the necessary information to improve harvest procedures (reduce cost and improve working efficiency). These models can help analysts, designers, and developers understand the behavior of the objects in a system (dynamic details of the behaviors). Further, we describe how to use these analyses and the modeling of all activities involved in the harvesting process to develop advanced automated data monitoring systems. In the systems we are developing a multitude of sensors are used (including digital weighing scale, RFID reader, computational unit, wearable datalogger), and new harvest methodologies are being developed, to track essential activities and information for traceability and improved spatial management (i.e., precision agriculture). Finally, an algorithm (i.e., software) will be presented that simulates the harvesting procedure and compare potential modifications under predefined controlled conditions. The proposed modeling process and examples serve as a referential model for harvest planning and a means for optimizing processes without the need for repeated field experiments.