Publikation
Real-Time Smart Farming Services
Wolfgang Maaß; Iaroslav Shcherbatyi; Sven Marquardt; Arndt Kritzner; Benedikt Moser
In: International Conference on Agricultural Engineering LAND.TECHNIK 2017. International Conference on Agricultural Engineering (AgEng-2017), November 10-11, Hannover, Germany, VDI, 2017.
Zusammenfassung
Agriculture resembles general production processes in many respects by being “production on the field”. Therefore, it is straightforward to apply concepts known by Industrie 4.0 to agricultural environments. In contrast to, for instance, automotive production, agricultural machines are constantly moving while products are rather static during growth phases. Harvesting and logistic processes increase the complexity by moving machines and moving crop. In this highly dynamic environment, sensor data is increasingly collected for any kind of signal, such as machine data (e.g., oil pressure, speed of rotation) and supervision data (e.g., video signals). Data flows in as packages or streams via communication protocols as defined by ISOBUS. Using this increasing amount of data for making decisions in real-time is currently a challenging task.In this paper, we introduce a framework for processing agricultural data by leveraging onboard computational devices in combination with cloud infrastructures. The concept of smart services is introduced as a means for processing data and providing services to various stakeholders. Special emphasis is given to real-time decision-making. The framework is derived from the more general acatech reference model for the Smart Services Welt initiative. Smart service environments are based on generic and domain-specific softwarebased services. This gives rise to a software market to which incumbents and new entrants will provide software-based services. Of particular interest are data analytical services based on various Artificial Intelligence methods.
The proposed RESFAST framework is exemplified by a potato harvesting process. Weintroduce the concept of a nPotato that senses physical impacts during the harvesting process. This data is analyzed locally on a harvesting machine in real-time. Impacts are categorized and accumulated over time. Farmers are informed about the current accumulated impact so that he/she can act accordingly and adjust machine configurations and driving speed. Furthermore economic forecasting services are used for making market price predictions that, in turn, are combined with impact analyses. Together, farmers receive information about the current status and the predicted return-on-investment. It is shown how data and results are used by cloud services for analyzing across different dimensions, suchas geographical areas, time, and crop types.