Vodafone, TomTom partner to create real-time traffic data network

Everyone who's ever been stuck in traffic always would love to know exactly what the road conditions are like. Sure there's those news radio stations that interrupt every three minutes to tell you how the freeway you're on is totally backed up, but those often don't tell you how to route yourself around the problem. TomTom and Vodafone have just partnered to create a new type of commercial traffic data system based on thousands of mobile phones that will describe traffic conditions in real-time. The idea is that by using the regular signaling information between the handset and the base station, the location and speed of the handset can be determined at any given time. Combine that information across a region among thousands of drivers who become data points, and a picture emerges of how backed a given freeway really is -- enabling TomTom to provide detours much more quickly than previously possible. The program is set to be launched in the Netherlands in the second half of 2007 -- so for all you folks who commute into the Dutch metropolises of Amsterdam, Rotterdam, The Hague and Utrecht, you may want to renew your Vodafone subscription pronto.

[Via Reg Hardware]

[Ben Miller @ Oct 28th 2006 1:17AM]

This approach seems much less capital intensive than the existing approach. It would also allow integrations of pedestrians, bicycles, motorcycyles, tuktuks and other more-than-a-motorcycle-bult-less-than-a-car vehicles into a overall trasport-congestion information system. The lower capital costs and the recording of the many types of users on the roads in places like China and India may make it particularly valuable in these markets.
http://www.ben-miller.com

[Razvan Dragomirescu @ Oct 28th 2006 6:42AM]

I can see a few obvious problems with this system: how do they differentiate between various types of vehicles? Ok, if you're on a highway you're probably not on a bike but to do traffic management in urban areas you would need to somehow identify the types of vehicles and classify them by speed and traffic patterns (e.g. taxis go fast but sometimes stop to pick up fares or wait for a client, bikes go slow but that's not a sign of congestion, pedestrians go even slower but that again is no sign of congestion. If I park my car and start walking, how do they tell that I got out of the car and it's not my car moving at 1mph? In an urban environment there might be more pedestrians than cars in certain areas and I doubt their system is accurate enough to detect that they are on the sidewalk and not on the actual street. But I guess I'll just wait and see, maybe they've already considered these issues.

[Victor Trac @ Oct 28th 2006 9:20AM]

I had this exact idea in 2004, and presented it as part of a business/engineering entrepreneurship class I had. The professors said that my plan was the most viable and most marketable and suggested I pursue it. Two and a half years later, looks like Vodafone and TomTom beat me to it. :)

[Ben Miller @ Oct 28th 2006 11:16AM]

Razvan Dragomirescu - search on "Alex Pentland." I don't really understand principal components analysis, but I take it that this technique (along with Branched Markov Chains - don't ask me!) offers a computationally efficient way around some of your issues.

[Bo Gus @ Oct 29th 2006 5:40PM]

This has got to be one of the smartest and most original ideas in a long time. I applaud the inventors and anybody else who had thought of it before I even got a chance to, including #3 - Victor Trac.

Implementation might be a little tricky, given the different types of cellphone users, but a search through academia, and I'm sure there are enough algorithms out there to make it work. And if cellphone companies started sharing this data with each other, even more accuracy.

INGENIOUS!!!