Skip to Main Content
The paper refers to intelligent industrial automation. The objective is to present key elements and methods for best practice, as well as some results obtained. The first part presents an ontology for automated cognition (cognitics), where, based on information and time, the main cognitive concepts, including those of complexity, knowledge, expertise, learning, intelligence abstraction, and concretization are rigorously defined, along with corresponding metrics and specific units. Among important conclusions at this point are the fact that reality is much too complex to be approached better than through much simplified models, in very restricted contexts. Another conclusion is the necessity to be focused on goal. Extensions are made here for group behavior. The second part briefly presents a basic law governing the choice of overall control architecture: achievable performance level of control system in terms of agility, relative to process dynamics, dictates the type of approaches which is suitable, in a spectrum which ranges from simple threshold-based switching, to classical closed-loop calculus (PID, state space multivariable systems, etc.), up to "impossible" cases where additional controllers must be considered, leading to cascaded, hierarchical control structures. For complex cases such as latter ones, new tools and methodologies must be designed, as is typical in O3NEIDA initiative, at least for software components. Finally, a large part of the paper presents a case study, a mobile robot, i.e. an embedded autonomous system with distributed, networked control, featuring industry-grade components, designed with the main goal of robust functionality. The case illustrates several of the concepts introduced earlier in the paper.