Design and Implementation of a Room Thermostat Using an Agent-based Approach

A large number of control systems operate in the real-world and are expected to deal with the range of changing conditions of the physical process that is being controlled. It may be a laborious and difficult task to design one controller that fits all operating conditions. In reality, several controllers, each suitable for a particular operating regime, are often combined in order to solve the problem. This paper describes a method that facilitates the design and implementation of these so-called multi-controller systems. It is based on the concept of an autonomous agent, which has recently been introduced in the field of artificial intelligence. The method enables the use of heterogeneous control algorithms and integrating techniques. Furthermore, it allows for an incremental design of the control system, which is attractive from a designer's point of view. The proposed framework is applied to the design and implementation of an ‘intelligent’ room thermostat in order to illustrate the applicability and usefulness of this method.