A Digital Twin of the Social-Ecological System Urban Beekeeping

We describe the system design and setup of our digital twin of the social-ecological system urban beekeeping, with the aim to support agroecological methods in urban agriculture. The physical space consists of the bee populations, their beekeepers who are part of a beekeeping community, non-beekeepers who consume honey, organisational actors shaping rules and regulations and the environment. The virtual space is a multi-agent model, where autonomous agents can take actions and make decisions in partially observed Markov processes. To tie the physical and the virtual space, we embedded bee hives in an IoT environment and implemented an online documentation tool as a web application, where beekeepers take short notes about their work and observations. Bee hives are equipped with sensors, such as humidity, pressure and temperature sensors and a scale. Additionally, we pull data from the German weather service (Deutscher Wetter Dienst, DWD). In our system architecture, multiple levels on data fusion are performed, beginning with raw data quality estimation and sensor failure detection. On higher levels, states of entities are estimated, such as the health of a bee colony, and assessment made whether a state is normal or to be considered an anomaly. Finally on the highest level, we deal with the desires of our agents, how actions should be chosen in order to achieve or maintain desirable and rewarding world states. We hope to be able to refine our digital twin into a decision support tool for small-scale (bee) farmers and communal political actors that helps to reach desirable world states by predicting and simulating the effects of actions within the complex system of urban beekeeping.