Estimating an operator’s cognitive state in real time : a user modeling approach

This paper presents a Cognitive Task Load (CTL) model designed to keep track of an operator’s mental workload, both quantitatively (amount of workload) and qualitatively (cognitive state). By integrating this model in a (semi-)autonomous robot, robot’s level of automation and user interface can be attuned to the operator’s state (Every second, the CTL-model updates a diagnosis of the operator’s cognitive state). The CTL-model’s predictions were tested in an Urban Search And Rescue (USAR) setting. Tasks could be well-monitored automatically, but the test showed insufficient workload variations to validate the model. This indicates that participants should be subjected to more ”high-pressure” conditions in future trials. These results also constitute evidence that adaptation to an operator’s mental workload is most effective when the operator’s ability to manage his own workload is restricted by the urgency of the tasks