In my thesis, I present a framework that could form the base of an intelligent greenhouse controlling system that empowers BDI (Belief, Desire, Intention) agents written in the AgentSpeak language for the Jason interpreter, to sense and control various aspects of a model greenhouse. These agents can then be organized in MASes (Multi Agent Systems) to harness their synergistic properties and facilitate the decomposition of the problem. I attempted to make the solution as generic as possible, hoping that the created software components would be useful not only for the model greenhouse but also for other environments with sensors and actuators.
First I describe the available technologies, the used tools and hardware, detailing the individual choices made for and against each item. This is followed by the description of the theoretical and practical operation of the realized framework, including the decisions made during the development.
Then, the operation of the completed system is demonstrated with the help of a few simple test problems and their solutions on the model greenhouse in the laboratory. The test problems are described and the respective agent societies are created to resolve them, in order to show how the completed system can bridge the architectural gap between the functional logic represented by the agents running in the Jason system and the sensors and actuators controlled by the existing programming in the microcontroller of the greenhouse model.
My conclusion serves as a review providing some possible and advised development directions for an industrial application or further studies and research.