Modeling, Simulation, and Realization of Cognitive Technical Systems

Abstract

This thesis presents a novel approach for the modeling, simulation, and realization of Cognitive Technical Systems. In contrast to other approaches, in this thesis, the structure and dynamic of the real world is initially formalized my means of an intermediate level instead of implementing a technical model directly. Furthermore, human cognition is investigated in an integrated manner and based on experiments with a mobile robot, as an example for a complex technical system. The formal description of human interaction and cognition is realized by Situation-Operator-Modeling (SOM), which can be implemented technically by patterns of high-level Petri Nets. With the state space of a SOM-based Petri Net, Human-Machine-Interaction can be analyzed, e.g., in order to detect human errors automatically. Furthermore, several cognitive functions, like planning, execution, perception, and learning, can be simulated. The different cognitive functions and related representations, which are all based on the same methodical background, are combined within an integrated cognitive architecture. Only the interplay among several functions and a novel kind of knowledge structuring, which contributes significantly to reduce the complexity of the real world, enable the realization of human-like behavior for technical systems. The system's capability to establish and to refine goal-directed behavior from interaction with the environment, also if no system-specific initial knowledge is available, is demonstrated by experiments with a mobile robot interacting within a dynamic office environment. An additional value of this thesis for further research is especially given by the proposed generic approach for modeling, simulation, and analysis of Human-Machine-Interaction. Moreover, the formal description and implementation of the cognitive functions, the developed knowledge structuring, and the cognitive architecture may be applied to arbitrary kind of technical systems.