MoDyS-Project KOGSIT

Operator modelling using cognitive architectures

The use of cognitive architectures to support the design and implementation of complex and dynamic human-machine-systems promises reduced cost and better functionality compared to building individual micro-theories. Current cognitive architectures like ACT-R or SOAR are not sufficiently equipped to support engineering complex and dynamic human-machine-systems. The goal of this project is to identify requirements and propose modifications to current cognitve architectures.

Building upon the experiences of the EnCoRe and FoG projects that aimed to build a cognitive model (Niessen & Eyferth 2001) of experienced en-route air traffic controllers with the cognitive architecture ACT-R, we encountered that current cognitive architectures are not suited for applied modelling in complex and dynamic human-machine-systems for engineering purposes (Jürgensohn et al. 2000).

Cognitive architectures were designed to cope with static problems in closed worlds but provide a rich set of mechnisms like learning or capacity and structure restrictions to the working memory that individual models can be build upon without additional efforts. Additionally current cognitive architectures are not yet very satisfying from a software engineering point of view. Support for building and maintaining cognitive models is not state of the art because environments coming from cognitive psychology are meant as tools for research where the underlying or tested theories are in focus. The deployment of such tools in engineering with cost oriented enviromental restrictions seems not to be anticipated originally.

Different case studies were used to gain experience and reveal problems of current cognitive architectures for applied modeling in dynamic human-machine systems. Modeling projects were conducted for tasks in process control in chemical plants and air traffic control. It turned out that ACT-R/PM is the most suited candidate for a cognitive architecture for applied cognitive modelling in human-machine systems. We developed and applied the following enhancements to ACT-R/PM:

ACT-R/PM enhancement for interfacing cognitive models with external technical environments via TCP/IP. Uses the visual and manual subsystem of PM for accessing information in the environment. Includes a GUI modelling tool for reuse of parts of models and a library.

Developed upon the results of "Operators' Timing in Human-Machine Systems". Additional buffer for perception of time durations in ACT-R/PM. Supports prospective as well as retrospective time duration estimation. Accounts for distorted time perception under different cognitve workload condition.

A framework for ACT-R/PM to process and execute action plans e.g. standard operating procedures. Plans are stored as declarative memory elements that can have to be retrieved from WM, they can be confused and forgotten. Execution of a plan influences the activation of it and its elements (steps). Includes a conception for compilation from task analysis to fine grained cognitive models. Does not yet deal with conditinals, interation, etc.

• Urbas, L. & Leuchter, S. (in prep.). Model Based Analysis and Design of Human-Machine Interaction with agimap. KI [artificial intelligence] special issue Dialogsysteme [dialogue systems].
• Kindsmüller, M.C., Leuchter, S., Schulze-Kissing, D. & Urbas, L. (2004). Modellierung und Simulation menschlichen Verhaltens als Methode der Mensch-Maschine-System-Forschung. MMI-Interaktiv, Nr. 7, Juni 2004, 4-16.
• Leuchter, S. (2004). Human-Machine Systems - An Interdisciplinary Research Experience. Japan Society for the Promotion of the Sciences (JSPS). Tokyo: 23.08.2004.
• Leuchter, S. (2004). Human performance modeling applied in the nuclear power plant domain. Project report JSPS Summer Program. Tokyo: Japan Society for the Promotion of the Sciences (JSPS).
• Leuchter, S. (2004). Applied Modelling of Mental Processes With the Cognitive Architecture ACT-R/PM and agimap. Research Committee Human-Machine Systems, Atomic Energy Society of Japan. The University of Tokyo: 06.08.2004.
• Leuchter, S., Kindsmüller, M.C., Schulze-Kissing, D. & Urbas, L. (2004). MMI-interaktiv, special issue Modellierung und Simulation in Mensch-Maschine-Systemen.
• Leuchter, S., Kindsmüller, M.C., Schulze-Kissing, D. & Urbas, L. (2004). Editorial zum Themenheft Modellierung und Simulation in Mensch-Maschine-Systemen. MMI-Interaktiv, Nr. 7, Juni 2004, 1-3.
• Leuchter, S. & Urbas, L. (2004). A Framework for Adaptive Human-Machine-Systems Based on Human Performance Models. GI-Fachgruppe Adaptivität und Benutzermodellierung in interaktiven Softwaresystemen (ABIS). Workshopwoche Lernen, Wissensentdeckung und Adaptivität, Berlin: 04.-06.10.2004.
• Leuchter, S. & Urbas, L. (2004). Extending ACT-R for modeling dynamics and timing for operating human-machine systems. MMI-Interaktiv, Nr. 7, Juni 2004, 38-46.
• Leuchter, S. & Urbas, L. (2004). Useware Engineering mit kognitiven Architekturen. atp - Automatisierungstechnische Praxis 46, (9) 68-72.
• Leuchter, S., & Urbas, L. (2004). Useware Engineering mit kognitiven Architekturen. Useware 2004. Nutzergerechte Gestaltung technischer Systeme, Tagung Darmstadt, 22. und 23. Juni 2004 (pp. 39-46). Düsseldorf: VDI-Verlag (VDI-Berichte; 1837).
• Leuchter, S. & Urbas, L. (2003). Modeling dynamics and timing for operating human-machine systems. In F. Detje, D. Dörner, & H. Schaub (Eds.), The Logic of Cognitive Systems. Proceedings of the Fifth International Conference on Cognitive Modeling (pp. 279-280). Bamberg: Universitäts-Verlag.
• Kindsmüller, M. C., Leuchter, S. & Urbas, L. (2002). Arbeitskreis: Einsatz von Modellierung und Simulation in Mensch-Maschine-Systemen - Transparenz gestalten. In R. Marzi, V. Karavezyris, H.-H. Erbe & K.-P. Timpe (Eds.), Bedienen und Verstehen. 4. Berliner Werkstatt Mensch-Maschine-Systeme. 10.-12. Oktober 2001 (pp. 318-329). Düsseldorf: VDI-Verlag (Fortschritt-Berichte, Reihe 22; 8).
• Leuchter, S., Schulze-Kissing, D. & Urbas, L. (2002): Architekturgebundene Modellierung zeitgesteuerten Verhaltens in Mensch-Maschine-Systemen. In E. van der Meer, H. Hagendorf, R. Beyer, F. Krüger, A. Nuthmann, S. Schulz (Hrsg.), 43. Kongress der Deutschen Gesellschaft für Psychologie. Humboldt Universität zu Berlin: 22.-26. September 2002. Programm, Abstracts. (p. 164; PO0331). Lengerich: Pabst.
• Jürgensohn, T., Niessen, C. & Leuchter, S. (2000). Bedienermodellierung: Beispiele. In: K-P. Timpe, T. Jürgensohn & H. Kolrep (Hrsg.), Mensch-Maschine Systemtechnik. Konzepte, Modellierung, Gestaltung, Evaluation. Düsseldorf: Symposion, pp. 149-177.
• Niessen, C., Leuchter, S. & Eyferth, K. (1998). A psychological model of air traffic control and its implementation. In: F.E. Ritter & R.M. Young (eds), Proceedings of the Second European Conference on Cognitive Modelling (ECCM-98). Nottingham, UK, Apr 1-4, 1998. Nottingham: Nottingham University Press, pp. 104-111.
• Leuchter, S., Niessen, C., Eyferth, K. & Bierwagen, T. (1997). Modelling mental Processes of experienced operators during control of a dynamic man-machine system. In BB. Borys, G. Johannsen, C. Wittenberg & G. Strätz (eds), Proceedings of the XVI. European Annual Conference on Human Decision Making and Manual Control, pp. 268-276. Kassel, Dec 9-11, 1997. Laborbericht IMAT-MMS - 19, Kassel, Germany.