Reduction of Vulnerable Road User accidents in urban intersections: Needs and challenges in designing Advanced Driver Assistance Systems

Advanced driver assistance systems (ADAS) can be used to prevent accidents, or to reduce their severity. It is essential to determine what functional requirements such systems should fulfill to meet drivers’ support needs. An understanding of the underlying contributing factors and context in which the accidents occur is therefore crucial. One aim of this thesis is to identify drivers’ support needs in accidents involving vulnerable road users (VRUs) in urban intersections. Another aim is to identify the most promising ADAS for this accident type and to derive functional sensor, collision detection, and human-machine interface (HMI) requirements. A third aim is to develop an ADAS concept based on these requirements. Microscopic and macroscopic accident data were analyzed. The microscopic data, obtained from the European project SafetyNet, consisted of causation charts describing contributing factors for 60 accidents. These charts have been compiled by means of the SafetyNet Accident Analysis System (SNACS). This thesis aggregated the individual causation charts for the drivers. The macroscopic data, obtained from the Swedish national accident database STRADA, consisted of 9702 accidents. The results revealed that the most frequent contributing factor was failure to observe the VRUs. This was mostly due to reduced visibility, reduced awareness, and/or insufficient comprehension. An ADAS should therefore help the drivers to notice the VRUs and enhance their ability to interpret the development of events in the near future. Such a system should include a combination of imminent and cautionary collision warnings, with additional support in the form of information about intersection geometry and traffic regulations. The warnings should preferably be deployed via in-vehicle HMI and according to the likelihood of accident risk. To enable timely warnings, it may be necessary to predict road user intentions approximately 4 seconds ahead. The study also showed that the system must be able to operate under a variety of road, weather, and light conditions. It should have the capacity to support drivers when their view is obstructed by physical objects. To this end, it is recommended that onboard sensors be complemented by cooperative infrastructure-car systems. Consequently, an ADAS concept utilizing a vision based VRU detection system in the infrastructure is proposed. The VRU position and velocity were continuously broadcast to the cars in the vicinity. The cars used global positioning systems to determine their own position and velocity. Based on these data, each car’s system estimated its own collision risk with each VRU. If this risk was high, information about the intersection and a cautionary warning were issued to the driver via an in-vehicle HMI. An initial evaluation of this conceptual system indicated that several technical factors and human aspects need further investigation and development. These include mainly detection and tracking of road users as well as prediction of their intentions.