In this work, motivated by the precision agriculture (PA) paradigm, we address the problem of managing hazelnut suckering plants on a per-plant basis in a large-scale orchard. Suckering plants, or shortly, suckers, are basal shoots that grow at the base of a tree and compete with the tree itself for nutrients and water. Generally, in large-scale orchards, suckers are treated with the application of herbicide through spraying tractors that continuously spray the crops while navigating the whole orchard. This approach however does not consider the individual needs of each plant and it is definitely not environmentally-friendly since a lot of unnecessary solution is being drained in the soil. For this reason, we propose a novel fully autonomous sucker management architecture that is able to detect the presence of suckers for each plant, by relying on a You Only Look Once (YOLO)-based recognition system, reconstruct them in three-dimension and estimate the amount of herbicide solution needed for the specific plant, based on a data-driven approach. The herbicide solution is applied using a ground robot equipped with an RGB-D camera and a spraying system. This approach allows to significantly reduce pollution and waste. Experimental results both for individual components and for the entire architecture in a real-world (1:1 scale) hazelnut orchard located in Caprarola, Italy, are provided to corroborate the proposed architecture.
An autonomous spraying robot architecture for sucker management in large-scale hazelnut orchards
2023-01-01
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
DDC: | 629 |