Development of an Advanced Driver Assistance System for Pedestrian Collision Avoidance in Urban Intersections

In recent years, the introduction of active and passive safety systems into modern cars has contributed to a high standard of safety for the passengers. However, vulnerable road users such as pedestrians and cyclists still face a considerably higher risk of severe injuries, especially in urban traffic situations. In this way, the number of total fatalities in traffic crashes in the USA decreased from 42836 to 32719 between the years 2004 and 2013, but the number of pedestrian fatalities remained constant around 4500. But fortunately, due to new technologies such as vehicle-to-infrastructure (V2I) and vehicleto- vehicle communication (V2V), networked vehicles can be provided with additional sensor information about their environment, which offers a great potential for advanced driver assistance systems. That is the motivation of this thesis, which is about developing a functional system for pedestrian collision avoidance in urban intersections, assuming that precise information about the position and speed of the pedestrian is available. So the main objective of the work is to detect the possible future collisions from this information. Moreover, as it is necessary for validating the detection methods, an automatic braking system is also considered. The major output of this thesis is the implementation of one system which, in the simulation, really detects impending collisions, alerts the driver about them and truly avoids them by activating an automatic braking system