Model-based design of an embedded vision application:: A field report

This paper contains a field report on using model-based design techniques for developing embedded vision applications. A lane border detection algorithm was chosen as target application for testing the design approach. The algorithm detects the lane borders, which marks the drivable area in front of a vehicle. It is an algorithm for an autonomous vehicle which participated in the Defense Advanced Research Projects Agency (DARPA) Grand Challenge 2005. The algorithm was processed on an embedded stereo vision system which was mounted on that vehicle. This paper describes the assembly of the vehicle, the principles of model-based design, the development of the lane border detection algorithm and its functional behavior. Furthermore, it presents evaluation results and our experiences with this design approach.