A cognitive agent architecture for feedback control scheme design

We present a novel architecture for the design of feedback control schemes, aiming to automate the cognitive process performed by human experts when designing control schemes for certain systems. The work starts with the identification of types of cyber-physical modules participating in a feedback control scheme. Each module is defined as a functional mapping between inputs-vectors and an output vector. The inputs and outputs are then mapped to a semantic space formed by linguistic variables that model the expert knowledge. These semantic annotations of the modules are then exploited by a Cognitive Agent with semantic reasoning capabilities, to achieve the online configuration of a feedback control scheme, given a set of specifications and a database of available modules' implementations. The adopted knowledge modelling and reasoning techniques are driven by past efforts of the World Wide Web Consortium on the semantic composition of Web services. The applicability of the method is tested via a paper-based simulation of a use-case from the smart buildings domain.