Abstract
The number of traffic accidents in Japan tends to decrease because of deployment of advanced driver assistance systems and active safety systems. However, accidents involving pedestrians and bicyclists account for a large number of fatalities, and further countermeasures to avoid those accidents are necessary. There are several cases where collisions with pedestrians and bicyclists cannot be avoided with the current specification of existing systems. In order to reduce the number of traffic accidents while walking or riding a bicycle, further improvement of driver assistance systems is required. This paper proposes a potential risk index called accumulated risk to predict theoretical avoidance behavior based on the relationship between the acceleration amplitude for planned trajectory and the collision risk. In order to obtain the reference acceleration for each driving scene, risk avoidance behaviors were formulated according to the predetermined patterns of acceleration in both longitudinal and lateral directions. The formulation of the avoidance behavior enables the calculation of acceleration command values to coordinate with other ADAS functions.
