Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.
Preliminary Approach for Real-time Mapping of Vineyards from an Autonomous Ground Robot
Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.orgCitation: 2016 ASABE Annual International Meeting 162457331.(doi:10.13031/aim.20162457331)
Authors: Veronica Saiz-Rubio, Francisco Rovira-Mas
Keywords: Agriculture Automation, Geostatistics, Information management, Precision Viticulture, Robotics, Specialty crops
Critical decisions on crop management, especially for the wine industry, are often made too late for a cost-efficient improvement of the final product. Most of the information gathered from vineyards is only available at harvesting time when limited changes can be introduced in the production chain, usually related to transportation, machinery, and winery logistics. The acquisition of reliable data along the entire growth period of plants results in better management choices as the sooner the information is available, the sooner effective decisions can be made. In a Precision Agriculture context, obtaining data on time and several times per season is becoming a necessity, where the ultimate expression of promptness leads to real time assessment of crop parameters. The VineRobot European project wants to respond to wine producers‘ desires of receiving real-time information on canopy growth and grape maturity. The robot is being developed for scouting vineyards autonomously and display data on-the-fly. This paper describes the real-time mapping approach displayed in the robot control screen, which, at present, is capable of showing on-the-fly data for localization and plant status with a sensor providing Nitrogen and Anthocyanins that is under development. Preliminary results applying fundamental Geostatistics to raw field data are presented with the final goal of automating the process of building maps right after each robot mission. To meet this objective, a method to deal with massive sampling while delivering near-truth maps is explored. Results indicate that a clustering technique applied to a grid of cells using a double crown with the median can be suitable for the automation of map smoothing in quasi-real-time, but more field tests incorporating robot-acquired data of Nitrogen and Anthocyanins will be needed to confirm such results.