Development Of A Methodology For Non-Intrusive Mental Workload Measurement In On-Road And Simulated Driving

The aim of the research was to develop the non-intrusive physiological measure of using human facial skin temperature change as an indicator of mental workload. The forehead and nose temperature were obtained via thermography from the participants who drove in a simulator driving environment and/or in instrumented car experiments. The NASA TLX and the Modified Cooper-Harper metrics were adopted to assess the subjective workload for the validation of the physiological measure. Three driving experiments were conducted in order to acquire the physiological response and the workload score for the performed tasks. Forehead temperature was very stable throughout the experiments. Nose temperature dropped significantly after the experimental drive for all conditions in simulator test. Experiment 1 (NASA TLX Group: N=10; MCH Group: N=14) used simulator driving with different terrains as loading tasks. Neither the significant difference of the subjective workload nor the temperature drop was detected between different terrain conditions. In experiment 2 (N=33), mental workload was increased in a controlled manner by the introduction of mental arithmetic tests to the primary simulated drive. The mental arithmetic test conditions provoked a significantly greater nose temperature drop and also a higher perceived workload than the conditions without the arithmetic test. A weak correlation between the nose temperature drop and the subjective workload metric was yielded from the experiments. In Experiment 3 (N=13), facial temperature response and subjective workload score were compared between the simulator test and on-road driving. Driving in the simulator resulted in higher subjective workload and greater nose temperature drop than in real-car driving. When participants perceived a higher workload for a task, their nose temperature exhibited a greater drop. A significant correlation between the nose temperature change and the subjective workload score was found. Actual or potential applications of this research include real-time and unobtrusive mental workload assessment for human-system interaction development