Collision avoidance in dynamic environments applied to autonomous vehicle guidance on the motorway

This paper discusses an approach to automatic vehicle guidance on a motorway with the intention of avoiding collisions. The autonomous vehicle should be able to manage the tasks of a driver, therefore it has to manage complex traffic situations in real time. Environment information is provided by several vision sensor modules and stored in a central dynamic database. A system view of the environment is generated by data fusion and data interpretation based on data stored in the dynamic database that represents the current scene. This system view is transformed into a riskmap representation which integrates information about the street, the relative position and speed of obstacles and traffic signs. The riskmap is an egocentric map of potentials reflecting the risk at a certain position in the environment. In order to achieve the "humanlike" behaviour, each riskmap is built according to a driver model and a vehicle model.