Advanced driver assistance systems (ADAS) are state of the art in modern vehicles (SAE level 1-2). They support the driver and improve thereby the vehicle safety during manual driving. In critical situations, collision avoidance systems warn the driver or trigger an autonomous emergency braking maneuver to mitigate or avoid a collision. Also, automated driving vehicles (SAE level 3+) must be able to avoid critical situations and must be more capable than currently available systems. During automated driving, the vehicle is responsible for the driving task instead of the driver. Therefore, safe automated driving requires robust algorithms to avoid collisions with other traffic participants in every situation, especially in critical situations with pedestrians and a limited perception ability. In this work, we investigate how automated driving vehicles can handle critical situations with pedestrians on multilane roads with an emergency braking or evasion maneuver. We focus in detail on very critical situations, where pedestrians are crossing behind an occluded area, e.g. from behind a parked car on the side of the road. In these critical situations, a collision avoidance system is not enough anymore because of the limited time-to-react. It is not acceptable that an automated driving vehicle passes obstacles very slowly. Therefore, a collision avoidance system is combined with a situation awareness planner to optimize the driving velocity. The situation awareness planner considers the sensor’s visibility and the capability of the collision avoidance system to provide a set of collision-free trajectories. This combination has the advantage that the vehicle does not need to pass objects on the side very conservative. We evaluate the approach rigorously on a set of well-defined scenarios from the Euro NCAP test protocol.
|Journal||SAE International Journal of Connected and Automated Vehicles|
|Publication status||Published - 2019|