Two startups traffic in freeway data

Escape Route
May 03, 2005
By Rafe Needleman
The Release 1.0 Monthly Report (CNET) www.release1-0.com

In nearly every car for sale in the US today, you can get a navigation system. But in the cars where the navigation systems are optional, the take rate on them is surprisingly small. At a recent Forrester conference on the auto business, I heard the industry's theory on this: It's because US drivers tend to drive mostly from home to work and back, with occasional excursions to soccer fields and malls on main roads they know. They don't see the value in spending upwards of $2000 on a navigation system.

On the other hand, traffic congestion is a major bummer for US drivers. If there were a system that could tell drivers which of their few known routes to work or home was the least congested at a given moment, they'd likely pay for that information.

The new Acura RL already has such a navigation system in it. An upcoming Cadillac will as well, followed, no doubt, by many more car models. Nav-with-traffic is coming to the aftermarket too, from Pioneer and other vendors. Traffic is the killer app for in-car navigation.

However, the traffic data available so far is limited. Existing traffic systems use road sensor data and that's collated by companies such as Navteq, and then sent via satellite radio to in-car receivers, (Acura uses XM satellite radio). Two interesting startups I've seen recently say they have ways to get much better-quality data into cars.

The first is Circumnav Networks, founded by Steve Wollenberg, who previously developed traffic information services for Windows CE devices. His company turns cars themselves into traffic data-collection devices. Circumnav hardware installed in the car records position and speed and then transmits that information to other cars - it's a social network of traffic data. The other cars' Circumnav units then make intelligent routing decisions based on the data. The prototype data-collecting device doubles as a dash-top traffic and navigation system.

Using cars as "floating probes," as such a system is called in traffic lingo, seems obvious. But getting the data from the cars/probes to other cars is a problem: Using cellular networks to transmit the data would be too expensive. Circumnav solves this by using WiFi as the uplink method. Cars will blast their data to other cars that they pass (called a "joust" data exchange). If there's a critical mass of cars that creates an effective ad-hoc mesh of collectors and repeaters, you'll end up with cars having extremely current information about traffic data that they care about.

The critical mass issue is the Big If. To solve this problem, Circumnav plans to install WiFi receivers along major thoroughfares, where traffic data will be collected from passing Circumnav-equipped cars, stored and broadcast out to other passing Circumnav users as they speed by. The company will also use the existing pager data network to send the existing highway flow data from today's sources (mostly departments of transportation) to cars using a Circumnav device, so even solo users of Circumnav will get some useful data.

Once there are enough Circumnav devices in use and sharing traffic data on all major commuting corridors, the company hopes to use this flow data to give users the option to let their in-car navigation units choose routes that are the quickest available.

That said, the system is based on unproven technical and financial models, and is highly dependent on having enough users to get the network effect. Wollenberg rightly hopes that his company will be able to get out of the hardware business and simply sell the hardware designs and algorithms to other navigation companies that already have inroads into the auto manufacturers and the aftermarket.

While Circumnav is built on gathering up-to-the-second traffic flow data from the cars in the thick of it, Seattle-based startup Inrix is basing its traffic data on predictions. Using Bayesian-based algorithms spun out of Microsoft labs just a month ago (and $6.1 million in funding from Venrock and August Capital), the Inrix model is to take existing public traffic data (from ground loops embedded in freeways, for example) as the starting point, and then predict flows based on history, season, weather reports and other known events that affect traffic, such as sporting events, fairs, school schedules and construction projects. The system will also collect incident data from highway patrols and police departments, and combine that with predictions on how long particular types of incidents take to clear.

CEO Bryan Mistele previously ran the automotive group at Microsoft and says his company is the sole beneficiary of Microsoft's traffic-predicting intellectual property. The system is currently being tested in Seattle.

Mistele wouldn't tell me about plans to equip users with Circumnav-like data collection and transmission equipment, although it's clear that he has plans for some sort of ad-hoc mesh data collection. He did let on that he's planning on finding a way to get data from other providers of floating probe traffic data, such as AtRoad and MobileAria, which collect data from fleet vehicles. Getting a hold of this information and then repackaging it for the general public presents a business development challenge, but it sounds no harder than the challenge Circumnav will have building out its own network of floating probes.

I like both these companies and I would not be surprised to see them working together, since the combination of floating probe data and traffic forecasting makes perfect sense. I would also not be surprised to see Inrix emerge as the senior partner in this relationship. Circumnav's social network for traffic is incredibly cool, but Inrix's predictive system, while based on complicated algorithms, doesn't require as much new ad-hoc infrastructure and would be useful sooner. Inrix's revenue structure is also based solely on reselling traffic data, which is a much cleaner business model.

Circumnav Networks
URL: http://www.circumnavnetworks.com
Interim CEO: Dan Kohn
Headquarters: Palo Alto, CA
Employees: 20
Founded: 2001
Market: traffic data
Funding: amount to-date not disclosed; raising B round now

Inrix
URL: http://www.inrix.com
CEO: Bryan Mistele
Headquarters: Redmond, WA
Employees: 12
Founded: 2004
Market: traffic data
Funding: $6.1 million in A round
Profitable: not yet

Rafe Needleman has been tracking cool technologies and startups for many years, writing for Red Herring, Business 2.0, AlwaysOn, and now Release 1.0. He is Editor of Business Buying Advice for CNET. E-mail Rafe with feedback or pitches for companies to cover.