Author Description

One of the most challenging tasks for modern computer systems is to recognize situations and perform actions in the real world -- a task that is handled easily by humans. Modern concepts for Artificial Intelligence still have serious problems when operating in such scenarios. In bionic approaches, models of human perception and cognitive processes are investigated in order to improve the performance of computational agents. A technically feasible and unitary model of these processes does not exist, however. This thesis introduces the computational framework ARSi09/Lang for the decision-making unit of an autonomous agent, which transfers the theory of psychoanalytical metapsychology into a technically feasible specification. Metapsychological concepts are not adapted but described technically with respect to a possible implementation. By applying a top-down design approach, the functional components of human decision processes according to metapsychology are identified and specified. The application of the developed ARSi09-framework probes its performance and results in a modular decision-making framework.