After a literature survey on workload and the means to measure it, a series of experiments was carried out, first on a driving simulator, and second on the road. The aim was to establish a method in order to measure workload, method at the same time sensitive (albe to detect small differences in the driving difficulties) and not intrusive (with no influence on the driving performance itself). Several indicators were tried and assessed. Several primary measures (speed, steering-wheel movements), a psychological measure (pulse rate), and a secondary task which considered in discriminating high or low sounds, the subject being instructed to offset his answer, that is to say to indicate if the signal before last was high or low, were selected. This task proved to be less sensitive and more intrusive on the road than on the driving simulator, but still remained usable, particularly when the results were combined with the primary and physiological measures. Another way to use the data consists in detecting the peaks of load, and in identifying on a video the situations and/or locations associated with them. This method reveals great differences between subjects. It makes possible to detect the situations yielding peaks of load, even if these situations are of short duration.
Abstract