Automatic Multimodal Cognitive Load Measurement (AMCLM).

摘要

This report summarizes the research activities, results of the user studies, and research accomplishments out of the AMCLM project in the past year. We investigated the validity of using speech formants and their fusion to measure cognitive load automatically. For the research on eye-activity based cognitive load measurement, we had examined various features, including blink latency, fixation time, saccade speed and pupil size. We further investigated the use of pupil size for automatic classification of cognitive load in different luminance conditions and under various emotional stimuli. All together, we had carried out four sets of user experiments to validate the research outcomes in a range of task scenarios, including Stroop test, computer- based basketball training, and mental arithmetic (summation) tasks.