Robot-generated Crop Maps for Decision-making in Vineyards

Abstract

<abstract> <b>Abstract. </b>The adoption of Precision Viticulture in Europe is occurring at a slower rate than in emergent wine production areas of Australia and America where large fields favor the incorporation of automated systems. In general, European vineyards are smaller in size and traditionally operated. The objective of the VineRobot project is the enhancement of vineyard management through the combination of robotics, precision farming, and information technology. This project aims at designing, developing, and deploying a novel agricultural robot endowed with non-invasive biosensors to map key parameters related to wine production, in particular vine growth from the content of Nitrogen in leaves, and the level of Anthocyanins in red grapes as a measure of ripeness, and therefore harvest readiness. This paper describes the mapping engine that will be implemented in the robot prototype for on-the-fly monitoring. The proposed maps will be constructed in real time as the robot scouts the vineyard, and have to permit the unambiguous location of measurements as well as facilitate data fusion of key parameters. To do so, data points are confined into regular squares (tiles) of homogeneous properties, in such a way that the robot will paint a grid as it progresses with tiles following the Northing-Easting orientation defined by the Local Tangent Plane coordinate system. The mapping approach was validated in field tests conducted in 2015, using the robot GPS receiver as positioning sensor. As the biosensor in charge of estimating Nitrogen and Anthocyanins levels was not ready for the experiments, the capabilities of the mapping approach were demonstrated with other parameters read from the GPS. The results showed that time-invariant standardized maps can be generated on-the-fly, and merging information from various sensors to elaborate management indices can be achieved with the proper discretization of space.