Thanks to new advances in automotive technology, active safety systems such as emergency brake assist are designed to avoid accidents or reduce their severity. The triggering of such systems is often based on drivers' behaviour. Their efficiency depends on the experimental or real data representativeness, and on the triggering criteria definition. According to the literature, few experiments have been carried out with "normal" drivers in "natural" or emergency situations in order to search for variables correlated to the intention and the needs of drivers. These data are necessary to define a "borderline" between emergency and natural driving to ensure that the system will be activated only if drivers need it. An experiment was conducted on open road. A hundred volunteers drove approximately 100 km. Traffic parameters and subjects' actions on the car's controls were recorded and synchronized with the video recordings. 14000 braking actions were analyzed. The results are useful for the design of active safety systems. For example, in some "natural" brakings, a fast accelerator pedal release, close to emergency situation actions, is relatively frequent on the road. A few of these actions are effectively linked to potentially dangerous situations. Taking into account other parameters, such as car's speed, distance to precedent car or drivers driving style, may increase the emergency braking recognition. For the covering abstract see ITRD E825082.
Samenvatting