Student Modelling for Operational Skill Training in Air Traffic Control

Current student modelling techniques cannot be applied directly to operational training on simulators in complex and dynamic worlds such as Air Traffic Control because of the lack of appropriate expert models and time constraints. The objective of this research is to present a student modelling approach for individualising the simulation training that (1) is specific to operational skill training, (2) can rely on a realistic expert model, and (3) can be used as a basis for a cognitive assessment of the learning. We are concerned here with the problem of assessing the student for operational skill training in dynamic and highly risky domains, such as Air Traffic Control (ATC) or nuclear plant operations. The aim of this assessment is to derive guidelines for defining the next training exercise in a simulation-based intelligent tutoring system (ITS). In most ITSs, the student is evaluated on his knowledge (declarative or procedural), not on his operational skills, which are the ability to use these two types of knowledge in a timely, accurate manner, and the ability to cope with dynamic problem solving situations involving multiple activities and goals. Little work has been done in the domain of student modelling for operational skill training. Since a complete model of the domain knowledge is not available for ATC operational training, student modelling techniques are not directly applicable. The training system we are aiming at keeps the human instructor in the loop and assists him in monitoring and evaluating the student and in designing follow-up exercises rather than replacing him totally.