Evaluating simulator-based training of skill-based control behavior using multimodal operator models

This paper describes a novel method for analyzing the training effectiveness for skill-based manual control tasks based on multimodal human operator models. For skill-based tracking tasks, it is known that the adopted human operator dynamics can be modeled accurately with multimodal human operator models. In this paper, estimated human operator model parameters are used to explicitly quantify the changes that occur in the operator´s use of visual and motion feedback during skill-acquisition and transfer. A quasi-transfer-of-training experiment is described, in which inexperienced participants were trained to perform an aircraft pitch attitude tracking task, either in a fixed-base or in moving-base simulator environment. After the training phase, the participants were transferred to the other simulator setting, to reveal possible transfer effects. Preliminary results from one participant in each experiment group indicate that the fitted models are successful in revealing the changes that occur in the multimodal manual control characteristics of the participants, and show that convergence to a final skill-based control strategy requires significant training. Furthermore, the presented results suggest that there might be limited direct transfer from training in a fixed-base environment to a moving-base environment.