Posted by Lutz Bendlin

We got an interesting article in the mail recently about traffic flow data capture. It provides some insights into the status quo of the US consumer telematics market. Read on and then let us know your opinion.

Why North America Is Slow to Develop Reliable Real-Time Traffic Data
By Connie Li, PhD
There was once a time when weather forecasters were blamed for misunderstanding what Mother Nature had up her sleeve.  Today, it is traffic reporters and other travel advisors.They just haven’t gotten the bumps out of their traffic data systems to provide the kind of accurate information U.S. drivers can depend on for their commutes and other travel chores.

As a result, the daily grind continues.  That doesn’t mean progress has eluded highway and urban planners and engineers.  Technology breakthroughs have sparked several innovations for providing more up-to-date travel information to auto and commercial truck drivers.  They involve better sensors for alerting authorities to congestions and other problems.  And they also comprise improved means for compiling traffic data, such as the use of Global Positioning Satellite (GPS) devices and mobile phones, and for delivering it to drivers, whether via a Web portal, a GPS system or a tech-savvy radio or cell phone.  

Yet despite progress, major improvements are still required. The traffic data provided hasn’t reached a high level of quality and it’s uncertain whether local, state and the federal government will deploy the resources and improve the infrastructure to make the kind of difference needed to reduce congestion and improve travel times and conditions.   

The data provided via these services has yet to achieve a stage of reliability that drivers can consistently count on across the country to deliver accurate travel information. Today’s solutions place too much emphasis on employing a limited number of data sources to produce real-time information.  It’s an approach that contributes to substantial data gaps and inaccurate information.

Commuting Times Lengthen
Lengthening commuting times over the last quarter century or so provide the clearest indicator of the increasing need for reliable traffic information.  The average commuting time in the U.S. has risen to 25.1 minutes one way in 2007 from 22.4 minutes in 1990 and 21.7 minutes in 1980, reports the Census Bureau. 

In major cities, those average times are much higher.  The worst commutes, according to data from the U.S. Census Bureau and the Texas Transportation Institute, are those that eat up the most hours and are the least reliable. Among the nation’s 75 largest metro areas, according to Forbes.com, the worst commutes are in Atlanta, Detroit and Miami as well as Orlando, Dallas, Birmingham and Raleigh, N.C.  Commuters there can blame bad traffic, insufficient infrastructure and drivers who resist carpools and public transportation.

The stark reality is that the typical Detroit commuter faces traffic delays of 54 hours a year, more than their counterparts in California’s San Bernardino and Riverside exurb communities, who log 49 hours of delays, or those in Chicago or Boston with 46 hours each.  Wherever, the daily grind is not pleasant.
In another recent survey, IBM’s “Commuter Pain Survey,” 27 percent of 4,000 commuters in 10 cities said at some point, they had turned around and gone home rather than face more traffic.  And nearly one in five (19%) said their commute had affected their performance severely at work or school. 

What Explains the Problem?
What explains why so called real-time traffic information is still unreliable?  It’s not that some communities and highways aren’t using more sophisticated road sensors; they are, including microwave radar, infrared sensors, ultrasonic detectors and passive acoustic devices that attach to bridges, overpasses and lighting structures. 

Still, such sensitive instruments cover less than seven percent of all major roads in the U.S. And it’s good to remember such “real-time” speed is really a prediction, due to reporting latency, equipment malfunction and the fact that for the most part speed is actually derived from data sensors primarily collect to measure capacity and volume on the road.

  
Even with advances in GPS and other probe data collection, speed calculation is not the only measure of advanced traffic information.  At the most extreme example, such as a bridge collapse or an emergency repair to close a highway, police may divert traffic for miles before the incident.  Intervening road sensors may nevertheless suggest that traffic is flowing freely since no cars are using the highway.

What’s needed is more sophisticated modeling engines that integrate a range of traffic impact data, which “know” when to reduce the weight such traffic models typically give sensors or GPS flow data.  Traffic authorities still need solid reports of incidents as soon as possible to provide ”near-time” traffic information about problem spots. 

For example, Westwood One, which serves the nation’s leading news & information radio stations with traffic reports every ten minutes, has an army of “tipsters”, listeners who call in traffic tie-ups and other traffic events when they see one – often before the police or other agencies are on the scene.  Descriptions of incidents are essential, because while speed data can show congestion developing, meaningful information requires knowing whether the cause is a major hazardous-materials spill or a minor fender-bender.   

No Silver Bullet
Traffic and travel information can’t wait for elusive “silver bullet” traffic data to be discovered and solve congestion headaches. The market has shown that one major breakthrough won’t suffice.  Instead, multiple sources of real-time traffic data are necessary to develop dynamic and meaningful real-time travel advisories.  And so the focus is on reducing the latency of data sourcing, improving the run time of modeling, using tailored data structures, efficient handling of local variables, hybrid approaches and parallel computing.

Many vendors and traffic centers rely on more and more sources for their traffic data.  But there’s yet another challenge, particularly in North America, that affects the quality of that data: the lack of data standards.  Traffic information does not have a Nielsen™ rating system, like that for television viewership, to measure accuracy and performance in a way acceptable across the markets.  

What’s missing is any degree of standardization of how drivers should get information and of the value of the technology itself.  Unlike in Japan and in certain European countries, government officials haven’t really sat down to determine how the U.S. is going to progress with the availability of real-time and reliable traffic information, especially considering the range of typical conditions across the country.

One reason is some important participants are on the sidelines.  Like auto manufacturers, they each follow their own paths when it comes to determining what, if any, GPS functionalities and other technologies are included in their vehicles.  Some manufacturers, of course, are much farther along than others and are using sophisticated systems.  But the challenges they face include planning cycles that require years to develop and supply new innovations. 

Some vehicle makers, smartly, are deciding they should be installing technology that can be upgraded easily and inexpensively when new advances come to market.  These may include the potential for enhanced communications to connect the vehicle to advanced data and information, while also upgrading the driver interface to simplify interaction and augment safety.

It’s All about Tech and Traffic Management
In North America, breakthrough technology advances will help add to the quality of traffic data.  Already, a number of companies are moving swiftly into the broad area of floating vehicle data to calculate road speeds.  In the US, cell probes, GPS probes and even tracking data off of cell phones are all being researched and evaluated.  Key variables being addressed include data sampling requirements, latency and “denoising” of extraneous data samples. 

Other new technologies also are helping improve real-time traffic information.  Analysis of data downloaded from GPS devices on commercial fleet trucks, buses and coaches supports modeling of real-time traffic and travel information and advisories while also enhancing route planning and scheduling. 

Cost effective communications technologies enable dissemination of increasingly valuable information via the Internet, mobile phones, roadway message signs as well as traditional media. 
Looking ahead, traffic experts anticipate that today’s social networking technology will further improve the delivery of accurate real-time traffic information. 

With the right software, device and user interface, drivers can quickly report traffic tie-ups, weather difficulties and other traffic impacts via their social networks.  In many ways, those radio station tipsters are doing so already.  The challenge is to aggregate, process and integrate their reports with the other sources of traffic impact data in the traffic information models.  

All of these promising developments point to the need for multiple data inputs to deliver accurate travel advisories and meaningful traffic information.  We estimate that even perfect road speed information  - whether from sensors, probes or tracking data – would still only deliver 50-60% of the data needed for reliable real-time advisories and travel time forecasts. 

Using only one traffic data source is like trying to predict the weather using only barometric pressure, or the dew point.  And like the weather forecaster who should remember to look out the window every once in a while, we recognize that even the best forecasts benefit from on-going “eyes on the road” to monitor data outputs.

U.S. Governmental Units Must Do More
Since technology first aided traffic in 1923 when a traffic signal was installed at a Cleveland intersection, governments have spent lots of money on high-tech traffic control – and lots more will be spent in the years ahead.  The Intelligent Transportation Systems, or ITS, industry estimates that some $209 billion will be spent in this first decade of the 21st century. Local agencies have been aided by a 17-year law that lets states use federal highway funds for something other than construction.

 
Many defense contractors also have helped; since the U.S. fought the Gulf War, they have designed technology that detects movements and then delivers that intelligence in real time to generals and other military officials who assess the information and act accordingly.  This technology has been well suited for monitoring urban traffic.

Still, local, state and federal governments just aren’t doing enough in this area and consumers don’t yet comprehend what a difference these technology advances and multiple data sources can make to their daily commutes and travel times.

Today’s weakening economy will present challenges in the short run, although the expectation of continuing high fuel costs is a compelling case for continuing investment in traffic data and information.  And it’s not just commuters who will benefit from improved traffic information. 

Considering the more advanced traffic and travel information available in many countries in the European Union and Asia, it is recognized that information technology to help commuters, goods and services reach to their destinations more efficiently has positive economic benefits.  And in a fiercely competitive global economy, those transportation advantages can make a great deal of difference. 

Connie Li, PhD, is co-founder, president and chief executive officer of TrafficCast International, Inc., whose Web-based software platform produces route-specific, real-time traffic information and travel-time forecasts for emerging applications that deliver such personalized information via Internet and wireless devices.   

Conclusion

The article claims that there is no silver bullet, but personally I would disagree. I believe that the cell data based speed model (as tested by TomTom and Vodafone) is close to the optimum. Of course this is just my view and you may have an entirely different opinion.