Design of Pedestrian target selection with Funnel Map for Pedestrian AEB System

Abstract

Recently, numerous vehicles have been installed with an autonomous emergency braking (AEB) system for protecting pedestrians. This system helps in avoiding or reducing accidents by alerting the driver and controlling the automatic brake actuator before an accident. Moreover, the European New Car Assessment Program (NCAP) has stipulated AEB pedestrian systems as a standard requirement since 2016. This paper presents pedestrian target selection using a funnel map for a pedestrian AEB system. The concept of target selection is based on crash probability calculations by comparing the pedestrian's predicted position and the current position to deduce the vehicle speed before an accident occurs. It is necessary to allow early breaking to avoid an accident. To determine the precise warning and brake timing, the warning distance is calculated using the vehicle and sensor fusion information. The pedestrian target selection algorithm is tested using a real vehicle on a test track in three different scenarios for the Euro NCAP using a pedestrian dummy authorized by the Euro NCAP. Upon comparing the results before and after the application of the proposed algorithm, the longitudinal distance is shown to have a maximum margin of 1.5 m, and the vehicle speed has a maximum reduction effect of 24.7 km/h. Test results show that the proposed pedestrian AEB system can avoid or mitigate an accident when the vehicle travels at speeds up to 40 km/h.