MoDyS-Project SAT

Model Based Situation Awareness Training

One of the key-skills of operators in many human-machine-systems is building and maintaining situation awareness i.e. to know the states and elements of the system, especially of important ones, their relevance to the current situation, and to know their future states. We propose a general strategy for training situation awareness skills for human-machine-systems that is based on the application of cognitive models.

A situation awareness training (SAT) strategy that is known to be suited to some human-machine-systems, esp. cockpits, is attention guiding (e.g. Gopher 1993): Control- or display elements of the environment are colour coded according to their current relevance to the situation using a task simulation . The trainee learns implicit strategies for detection of relevance and sharing his or her attention between multiple focusses.

The only proposed technique for detection of relevance in the simulation environment is context based state monitoring ( CBSM, Bass 1998). In CBSM the simulation environment is monitored by a state machine. Every action the operator takes is a transition from one state to another. Thus the state machine "knows" at every point of the simulation which state the task environment has. For training purposes every state has attached information on possible training interventions such as colouring a certain element on a display for SAT.

CBSM is only feasible if the extension of all potential states of the system is possible to enumerate and its size can be handled. Most dynamic and complex human-machine-systems are not suited for CBSM for this reason. E.g. in en-route air traffic control there are many possible actions the controller can take to avoid a pending conflict. Not only the actions, but also the time when to intervene is not fixed so the resulting traffic can not be known in advance.

For domains or tasks where CBSM can not be applied for this reason we propose state monitoring using simulations of cognitive models of experienced operators (Leuchter 2000, Leuchter & Jürgensohn 2000, 2001). The cognitive model runs parallelly to the task simulation that the trainee uses. It receives information from the simulation, builds and maintains its represenation of the current situation on the basis of it and draws inferences to anticipate future states or to detect special system states. The current situation is thus in the "mental model" ie. this representation (its DMEs: declarative memory elements) accessible and can be used to guide attention by colour coding in the task simulation.

A simulation training system with SAT was implemented for enroute air traffic control (Leuchter 2000, Leuchter & Jürgensohn 2000, 2001). The conception was extended with results from PIC4SAT and transfered to process control in chemical plants (Urbas & Leuchter 2002, Leuchter & Urbas 2002). Results of a evaluation study revealed directions for further developments (Bruder, Leuchter, Schulze-Kissing & Urbas 2002).

• Leuchter, S. & Urbas, L. (2004): Human Performance Modelling and Its Application in Operator Training. 14. Arbeitstreffen der GI-Fachgruppe Intelligente Lehr-/Lernsysteme auf der Modellierung 2004, Marburg: 25.03.2004.
• Bruder, C., Leuchter, S., Schulze-Kissing, D. & Urbas, L. (2002). Evaluation von Situation Awareness Training in der Flugsicherung. In M. Grandt & K.-P. Gärtner (Eds.), Situation Awareness in der Fahrzeug- und Prozessführung. (pp. 319-338). Bonn: Deutsche Gesellschaft für Luft- und Raumfahrt e.V. (DGLR Bericht; 2002-04).
• Leuchter, S. & Urbas, L. (2002). Simulation Based Situation Awareness Training for Control of Human-Machine-Systems. In Valery Petrushin, Piet Kommers, Kinshuk, & Ildar Galeev (Eds.), IEEE International Conference on Advanced Learning Technologies. Media and the Culture of Learning. Kazan, Russia: Sep 9-12, 2002 (pp. 34-39). Palmerston North, New Zealand: IEEE Learning Technology Task Force.
• Urbas, L. & Leuchter, S. (2002). Modellgestütztes situation awareness-Training für komplexe und dynamische Mensch-Maschine-Systeme. 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. 71-85). Düsseldorf: VDI-Verlag (Fortschritt-Berichte, Reihe 22; 8).
• Leuchter, S. & Jürgensohn, T. (2001). Situation Awareness-Training für Fluglotsenschüler. In: H. Oberquelle, R. Oppermann & J. Krause (Hrsg.), Mensch & Computer 2001. 1. Fachübergreifende Konferenz, Bad Honnef, März 5-8, 2001. Stuttgart: Teubner. pp. 437-438.
• Leuchter, S. & Jürgensohn, T. (2000). A tutoring system for air traffic control on the basis of a cognitive model. In: J.L. Alty (ed), Proceedings of the XVIII. European Annual Conference on Human Decision Making and Manual Control. Loughborough, UK, Oct 25-27, 1999. Group D Publications, pp. 275-281.
• Leuchter, S. (2000). Integration eines kognitiven Modells in ein Simulationssystem zum Training der Luftverkehrsüberwachung. Diplomarbeit am Fachbereich Informatik der Technischen Universität Berlin. MMI Interaktiv 3, Juni 2000.
• Urbas, L. (1999). Entwicklung und Realisierung einer Trainings- und Ausbildungsumgebung zur Schulung der Prozeßdynamik und des Anlagenbetriebs im Internet. Dissertation. Düsseldorf: VDI-Verlag (Fortschritt-Berichte VDI Reihe 19; Nr. 614).