Gavrila et al. 2000 : The authors discuss a vision-based driver assistance system implementing a sophisticated cruise control called Intelligent Stop&Go. They provide a framework for object recognition, proposing a two-stage approach: multi-clue object detection followed by intensity-based object classification. They furthermore describe a way to integrate various vision modules into a scalable multi-agent system. Recognition results were presented on traffic lights, traffic signs and pedestrians; these and other modules are running on-board the demonstrator vehicle UTA II. Though the authors have been fairly successful with the demonstrator UTA II so far, work remains on improving the performance of various vision modules and their integration. For the recognition tasks this mainly involves improving the ratio of correct classifications versus false positives. However, algorithmic advances represent only one part of the solution. The sensor considerably influences system performance and it is well known that CCD cameras lack the dynamic range that is necessary to operate in traffic under adverse lighting condition (i.e. allowing the camera to capture structure in shadowed areas when exposed to bright light). Here, CMOS camera technology can be of help. The authors are furthermore interested in improving performance by sensor fusion, e.g. combining vision with DGPS (Differential Global Positioning System), inertial systems, and digital maps. Finally, the telmatics solutions will undoubtedly provide added benefits to a sensor-based system.