Brief Survey of Cognitive Architectures

Abstract

Existing communication utilities, such as the ISO/OSI model and the associated automation pyramid, have limitations regarding the increased complexity of modern automation systems. The introduction of profiles for fieldbus systems, or field-area networks (FANs), was an important innovation. However, in the foreseeable future the number of FAN nodes in building automation systems is expected to increase drastically. And here the authors see an opportunity to revolutionize the operation of intelligent, autonomous systems based on FANs. The paper introduces a system based on bionic principles to process the information obtained from a large number of diverse sensors. By means of multilevel symbolization, the amount of information to be processed is substantially reduced. A symbolic processing model is introduced that enables the processing of real world information, creates a world representation, and evaluates scenarios that occur in this representation. Two applications involving human actions in a building automation environment are briefly discussed. It is argued that the use of internal symbolization leads to greater flexibility in the case of a large number of sensors, providing the ability to adapt to changing sensor inputs in an intelligent way.