Towards learning adaptive workload maps

One approach to mitigate the risks of driver distraction is to build an in-vehicle service manager component that is aware of the attentional requirements of the current and of upcoming traffic situations. This component will rely on technologies for personalized driver workload prediction, based on an enhanced digital map, and/or on sensors for physiological and behavioral workload correlates. In this report, we address first results of our approach towards the following questions: (1) According to our experiments, what method is best for online/predictive workload estimation? (2) Which sensors are most suitable? (3) How do physiological measurements and subjective rating correlate? (4) Which proportion of workload can be statically predicted (based on map features alone)? (5) How do workload patterns differ between drivers? (6) How dynamic is workload (how long does an influence persist)? and (7) Which factors (percentage) influence workload?.