Just a few random notes I picked up along the way:
Congestion is estimated to cost the U.S. economy $78 billion per year.
Twenty-seven percent of police-reported vehicular accidents are intersection collisions.
Using historical traffic data to project future traffice is about 80% accurate (way better odds than the stock market).
Intelligent vehicles that have the capability to communicate with infrastructure and each other would be analogous to data probes. An example given was if most vehicles in a given area started using their windshield wipers, that data could provide a real-time weather report.
One of the presenters, Dr. Luca Delgrossi of Mercedes-Benz, stated that safety is the priority when it comes to intelligent vehicles because the requirements are much more demanding compared to those for providing mobility and convenience functions. If the safety portion is achieved, the other uses will also become available.
More accurate map data would help reduce intersection collisions. So-called regular GPS can predict what road a vehicle is traveling on. More sophisticated GPS would enable pinpointing which lane a vehicle is in.
The difference in lifecycles between cars and cell phones is a huge challenge to bring them together. What's needed are flexible computing platforms in vehicles that can be upgraded to keep pace with changes in communication technology.
Improvements to vehicles by themselves are mostly controlled by the manufacturer. Improvements to the networks of vehicles is dependent on collaboration between multiple stakeholders and manufacturers, which makes the challenge far more complex. It's a classic chicken-and-egg problem.