Robots go automotive - the SPARC approach

This paper introduces a new concept for advanced driver assistance by means of a redundant architecture including all system components spanning from environment perception to vehicle controllers. The first part of this paper is an overview of the project framework and the research platforms. After that the elements of the architecture themselves will be described. The use of sensors and the fusion of their outputs will be presented. Different controllers will be used depending on the scenarios around the vehicle in order to provide a theoric solution. This solution will be downsized after that with a dynamic vehicle model to the feasible safe motion vectors. This list of motion vectors will be compared to the driver's command and will lead to the choose of his/her command or an other safe motion vector if the driver does not react convenient to the situation. The final part describes some preliminary results and concludes towards future work and research issues.