It’s one of the most dreaded questions of road-tripping parents everywhere: how much further? In the not-too-distant past, a vague estimation would have to do.
Quickly dividing distance by speed will get you part of the way there, but what about congestion? What about accidents? What about road construction?
Traffic is more than just an inconvenience on road trips – daily commutes are often unavoidable and costly. According to the Center for Economics and Business, traffic costs Americans $124 billion a year or roughly $1,700 a year per household.
Traffic conditions can change in an instant, often taking drivers by surprise. But more and more options are showing up to provide drivers with the information they need not only to know how much further they must go in their trip, but what alternate routes they can take to try to beat congestion. To earn driver trust, this kind of trip time information must be consistently accurate and available when the driver needs it.
The first step in the trip time generation process may also be the most important: accurate traffic detection. Traffic measurements determine the accuracy of the entire system. Expensive, complicated processing or delivery components won’t make up for inadequate detection.
Several different technologies are used to detect traffic and measure existing conditions. These typically fall into three categories: intrusive devices embedded in the roadway; non-intrusive devices which are typically installed above or to the side of the road; and in-vehicle devices, such as smartphones, which provide vehicle information passively.
For years, inductive loops buried in the road were the standard for data acquisition. Loops are very accurate, but suffer from many problems inherent with being intrusive, or in-ground, technologies. First, loops and other intrusive devices are much more difficult to repair or replace. Failures require the road to be dug up, causing lane closures that negatively impact traffic flow and place road workers at risk. These repairs and replacements can take anywhere from a few hours to a few days. According to a study performed by the New York Department of Transportation, at any given time 25 percent of all installed inductive loops are not operational due to their high failure rate.
At any given time 25 percent of all installed inductive loops are not operational due to their high failure rate.—NEW YORK DEPARTMENT OF TRANSPORTATION
Second, inductive loops are difficult to install properly. In fact, design and installation mistakes are the number one cause of loop failure, according to the US Department of Transportation’s Traffic Detector Handbook. Wire failure and sealant failure can also cause the systems to shut down. By its very nature, intrusive systems rely on the integrity of the road surface in order to work properly. Freeze/thaw cycles, road surface failure, construction work, even lane reconfiguration can disable loop systems.
Non-intrusive devices, such as radar and cameras, solve the problems inherent with loops by going above the road. They are typically easy to install, requiring no lane closures and with minimal time and manpower.
Additionally, in most cases, a single non-intrusive device can do the work of several in-pavement devices. For example, the SmartSensor HD is a radar traffic sensor that can accurately detect traffic in up to 250 feet and 22 lanes of traffic. Using a patented dual-beam radar design, HD is able to create a virtual speed trap that gathers an entire suite of data for trip times and other ITS applications. You would need at least 44 inductive loops, along with the extra time, manpower and failure points that come with it, to get the same coverage as one SmartSensor HD.
The SmartSensor HD is a radar traffic sensor that can accurately detect traffic in up to 250 feet and 22 lanes of traffic.
While non-intrusive devices tend to be more reliable than intrusive devices, the accuracy of the information gathered can vary wildly between devices. Detection needs to provide accurate information in all lighting and weather conditions and perform reliably without the need for constant repairs and maintenance. Care should be taken when choosing detection devices to ensure the foundation of trip time information, as well as any ITS application, is secure and reliable.
The newest player in the realm of traveler information comes via the explosion in popularity of smartphones. These systems use GPS tracking to not only provide directions but also trip times and traveler information. This information is curated from historical traffic data as well as real-time information made from tracking the position of smartphone users who drive. Most smartphones are, by default, tracked continuously by cell phone manufacturers. Trip time information is then sent back to the driver in the form of maps and GPS applications.
However, relying on smartphone tracking for trip times has several drawbacks. First and foremost, there aren’t enough phones to get accurate real-time information. According to the Pew Internet Project, only 58 percent of Americans have a smartphone. While that may go a long way toward creating estimations, almost half of all drivers remain unaccounted for, leaving large holes in data that make the information less accurate and inappropriate for planning purposes.
Safety is another concern. More and more states in the US are making cell phone use illegal while driving. Indeed, using cellphones while driving has been proven dangerous. The National Safety Council found that around one in four automobile accidents is caused by cell phone use and the Journal of Safety Research reports that hands-free devices are “rarely found to be better than using a handheld phone.”
While some cell phone manufacturers passively track all phones in cars, many services require a special application to be running in order to work. Some of these applications have gamification elements which reward drivers who actively input information such as the location of traffic jams, police officers and accidents, which could lead to more distraction and more accidents.
Finally, privacy is another concern when it comes to tracking the location of smartphone users – though that concern starts to disappear the younger people are. According to a survey from the University of Southern California’s Center for Digital Future, 56 percent of millennials, ages 18-34, are willing to have their location tracked if they get something out of it, like improved trip times or coupons for nearby businesses, compared to only 19 percent for those 35 and older.
There are still many questions as to how crowdsourcing will fit into the future of ITS, but it seems destined that crowdsourcing will play a part. Google has recently partnered with transportation agencies around the world in a pilot program where data will be traded to improve the overall user experience for the transportation agencies as well as for Waze, a crowd-sourced GPS and trip time application owned by Google. This partnership allows transportation agencies to supplement their data with the user-generated data from Waze, but it still requires agencies to have an accurate detection base to work from.
Typically, a transportation department will use several methods of data acquisition to create trip time information. Once the data is measured, it must be collected at a central location for processing and delivery. To facilitate real-time generation of accurate trip time estimates, a data collection system should meet several key criteria.
First, it should natively support as many measurement device types as possible. This gives agencies some flexibility to use the measurement ecosystem currently in place regardless of the measurement device.
Second, the system should use TCP/ IP as the collection protocol and should address potential communication problems by employing robust collection methods such as multiple collection retries and retrieving saved traffic data from a device once connectivity has been restored.
Third, the collected data should be stored in a relational database for easy access to current and past information. Since the data collection system will need to integrate with other systems, it should be easy to retrieve data from the database in modern data exchange formats.
Fourth, any system will eventually need to be expanded. System expansion should be straightforward so adding new detection or communication devices is a simple matter.
Finally, the system should maintain a constant uptime. Inevitably, computer hardware or software failures, as well as communication and measurement device problems will occur; if these are not dealt with promptly, the reliability of the resultant trip times will be compromised. An automated process should be employed to continuously check the connectivity to each measurement device. If problems are found, the system should generate regular reports that can be used as part of a maintenance program.
After the measured data is collected to a central location, it must be moved to the data processing component so trip time estimates can be generated. While newer systems may be fully integrated into a single system, typically systems are composed of several independent components that perform specific functions within the highway network. The data will likely need to be translated into a common format and interval before trip time generation can occur.
Most processing systems use one of three types of algorithms to estimate trip times. Currently, the most widely used algorithm estimates the “instantaneous” travel time – or the time it would take a vehicle to travel along the route if the speed remained constant. This requires accurate current speed measurements along the entire route; the advantage of this method is simplicity, but with changing congestion levels it can under- or over-estimate travel times.
The other two algorithm types attempt to predict how the speed will change during the time it takes a vehicle to travel along the route and then use this predicted speed to estimate the travel time. The two fundamental prediction methods are model-based and data-driven and both work best if all three traffic parameters (volume, speed and occupancy) are measured. A model-based approach uses a mathematical model for traffic flow driven by current traffic measurements and an estimate of the relative demand between the possible origin-destination (OD) pairs. Model-based methods work very well but it’s difficult to estimate the OD demand.
Meanwhile, a data-driven method uses historical data to train an algorithm that can then use current traffic measurements to predict future traffic flow. This method also works well but it requires training before it can be used. However, this method can be automated, whereas the OD demand estimation is typically a manual process.
Once trip time estimates are generated they are ready for delivery to the commuter. There are many different delivery methods available and each may require the data in different formats. Once the data is converted into the right format, it must be sent to the specific delivery method, a process that may be as simple as populating a database table or generating an XML file; or it may be as complicated as formatting an entire HTML page or providing the data for any number of smartphone or tablet applications.
No matter the delivery method, it is a process that requires the ability to easily integrate with other systems. The most reliable and most easily accessible methods include variable message signs, radio, phone and web; however, work is being done to implement more in-vehicle technologies, so system integration needs to be done in a way that addresses current technologies while allowing for easy additions in the future.
Growing strains on time and money means drivers are no longer satisfied with a vague answer to the question “How much further?” Without proper trip time knowledge and the ability to know how to avoid accidents and congestion, drivers all too often end up stuck in traffic, stressed out and wishing for better control.
Growing strains on time and money means drivers are no longer satisfied with a vague answer to the question “How much further?”
Trip times are one of the simplest ways to enable drivers to make informed decisions.
For effective congestion control, trip times must be accurate and available whenever a driver needs the information. To provide reliable trip times, agencies must take steps to ensure the integrity of their traffic data. By choosing measurement technology that provides accurate performance; by fully integrating the system for easy transfer of data from one component to the next; and by ensuring the system will be easy to expand in terms of increasing measurement coverage and delivery methods, agencies will give their customers the best chance possible to combat congestion.
