Modelling a multi agent-system for smart homes

Abstract:
In future smart homes, the number of household devices will be constantly increasing. To simplify their control, automation technology can be used. When integrating additional devices in open/closed loop control systems dedicated to the domotics area, however, the complexity of the overall system continues to grow. In addition, the energy efficiency of the underlying systems and their energy efficient operation are becoming increasingly important -- among other reasons due to economic considerations as well as a growing environmental consciousness of our society. For control tasks, modern home automation technology already provides a solid base. Unfortunately, the present systems are usually complex to configure and operate. Typically, necessary adjustments are once done during the commissioning phase and seldom changed afterwards. To guarantee that the systems will later on operate at optimum efficiency, an autonomous operation together with automatic re-adjustment is necessary.
In this work, a smart home system based on artificial intelligence is developed. The system design follows a multi-agent approach resting upon an established agent development methodology. Besides the domotics area (heating, ventilation, air conditioning, lighting/shading) it also takes into account various energy consumers and residents. Agents represent dedicated actors of the smart home, some act as avatars for their inhabitants. Together these agents aim at achieving a common goal: to operate the smart home as energy efficient as possible while keeping comfort for their residents in mind. Based on a detailed system specification, the system has been prototypically implemented and tested.