Applying the thrid and fourth dimension to precision agriculture in apple production

"In fruit production reliable and accurate information is needed for optimal orchard management. Due to an increase in farm size and high labour input, a grower has reduced information feedback in time and space from the orchard. On the other hand, more detailed information about the fruit production is required in the value chain to meet consumer demands and regulation requirements. From arable farming, we know that it is possible to gather spatial data and use this data in order to make management decisions. Therefore, the objective in our Fruit 4.0 project is to gain insight in apple production by gathering 3D and 4D data and improving data management. In order to improve current practices we quantify the most important parameters used in decision making during the growth season. The number of blossom clusters, fruit number and size, shoot length, leaf vitality, leaf quantity and fruit colour. All of these parameters will be determined on a sub-tree level, so a 3rd dimension (tree height) is required. Throughout the growing season time series of tree parameters such as canopy volume, height and vitality are determined, the development of these parameters in time is considered the 4th dimension. Sensors used for these purposes are a 2d-lidar system, a chlorophyll sensor and a RGB-d camera.
Apple trees produce variable numbers of flowers per year, in general too many, with a high variability. To reach an optimal number of flowers and fruits, chemical, mechanical or manual thinning is applied by the grower. To automate this process RGBd images are taken throughout the flowering period and at fruitlet stage combined with different classification techniques to estimate the number of blossoms and fruitlets per tree, in order to apply tree-specific thinning in future. At later stages of the growing season the number and size of fruits are determined to optimise yield estimations and the harvest time, so labour requirements can be better estimated and planned. A management system setup for growers will be presented in which the obtained sensor data can be visualised and can be used to create a precision application map for orchard sprayers
"