April 16, 2011

Routing podcast review

CNET Reporters' Roundtable had a very informative podcast about the state of car navigation technology. Here's the link


Craig Chapman, CTO of Inrix, gives a quick rundown of routing algorithms:
  • core algorithms unchanged for 20 years, like A*
  • first generation algorithms used static data
  • second generation incorporated incident avoidance
  • current third generation algorithms use real time information collected from cameras and fleet vehicles along with predictive algorithms
Real time info about traffic can come from:
  • radio stations and helicopters (more than 6 years ago)
  • city sensors in the roads (for about 20 cities)
  • road sensors, toll tag readers, radars, and, for Inrix, about 4 million contracted fleet vehicles
  • planned events that disturb traffic like sport games, conferences or parades
Waze relies less on fleet data, and more on citizen traffic, because fleet vehicles have different behavior than regular drivers. Google uses dumb probes form GPS enabled Android phones to collect their traffic data, though the quality is not as good as Inrix's.

Rafe Needleman asked whether in case of a jam, will the navigation systems direct all the cars to the same route and just move the jam to another place. A real time navigation system will not cause a jam, because the traffic will just normalize.

The best navigation system devices are now the iPhone and Android phones because they can get real time data. Personal navigation devices (the things on suction cups on your windshield) are going away, and legislation must be changed to allow using phones for navigation while driving. Ford, Audi, and Toyota will soon use Inrix data in their built in navigation systems.

The near future of routing will incorporate parking data. Some companies like Parking in Motion and Best Parking are already collecting data from parking garages. With parking meters going electronic, it may be possible that cars will interact with them and help us find quicker parking.