A DIY sensor kit, Gaussian Processes and a multi-agent system fused into a smart beekeeping assistant

摘要

The citizen science project BeeObserver provides open data which are recorded in and about honey bee colonies. Synchronized with sensor measurements, observations made by beekeepers are gathered through a web application. This application, the BOBApp, is also the interface which provides the beekeepers with access to the measured hive data. The sensors, the BOBApp and the overall system architecture is described in this paper. One of the motivations for implementation of the BeeObserver system is to be able to detect diseases or other precarious situations as early as possible and to inform the beekeeper about it in a constructive manner. To achieve this, measured data are fused with further data sources (such as open weather data) and evaluated using cognitively motivated algorithms. One of this precarious situations, food deprivation during winter, is exemplarily described in detail. To overcome the challenge of testing an assistance system within the comparatively slow dynamics of the living organisms, whose behaviour depends on the yearly rhythm of the seasons, we designed an agent-based model as a software-in-the loop testbed.