agriOpenLink: Semantic Services for Adaptive Processes in Livestock Farming

Abstract

Intelligent machinery and information and communication technologies (ICT) are crucial enablers for precision arable farming and automatized livestock husbandry. Technologies such as advanced robotics, sensor technology, and localization systems are being deployed to increase efficiency and reduce production costs based on efficient use of resources. Robots and sensors are vital sources of data that directly assist farmers in their work. For example, with the help of a localization application based on active ear tags a farmer can quickly identify any particular animal even within a large herd. On the other hand, farmers may have too many equipment to handle and operate, and they often need to manually input the very same data in different machines through different user interfaces (UI), and account for, and act upon data these machines offer them mostly for manual inspection. Accordingly, there is a growing need for a farm management platform that could provide a unified interface towards variety of different systems, as well as combine data from different sources and provide decision support to the farmer. This need is also motivated by the common awareness that the more data sources get integrated within the farm management decision support systems, the more optimal and sophisticated become the decisions or recommendations that such systems can offer. For example, in a dairy farming scenario an integrated system of a milking robot and an animal localization system with movement and rumination monitoring, can improve quality of collected milk, as well as the wellbeing and health of the animals. As the cost of designing decision support systems is generally high, it is essential that their architecture is open and future-proof to support integration of new innovative technologies as they emerge. Unfortunately, this vision of an open-platform is still far from reality. Approaches that are prominent today are based on single-vendor island solutions where new products can be integrated within the running system only if they are of the same vendor. On the other hand, to allow for a broader innovation adoption efficient plug-and-play of heterogeneous systems, processes and data is urgently needed. This paper presents a novel approach towards open interfaces and process models in agricultural production that is being developed within the project agriOpenLink. Our methodology is based on service and semantic technologies. To integrate heterogeneous devices we develop a common platform - a plugin server - and tools for plugin development. Each plugin publishes a number of semantic web services which offer data for integration. They run on the plugin server, and are formally described using the common ontology. The paper discusses our goals, methodology and results achieved so far, and illustrates the solution in the use case of dairy farming and livestock husbandry.