Jason Gao, a Ph.D. student at the MIT Computer Science and Artificial Intelligence Laboratory, and Li-Shiuan Peh, an MIT professor of electrical engineering and computer science, have an idea for eliminating gridlock. The two researchers hope to reduce traffic congestion by having cars communicate with each other via low-power, long-distance radios integrated into the drivers’ mobile phones or embedded in their vehicles.
Gao and Peh recently tested their concept, dubbed RoadRunner, using software simulation, as well as a real-world test involving 10 vehicles in Cambridge, Mass. For the simulation, they used Electronic Road Pricing data supplied by Singapore’s Land Transport Authority. For the live test, each driver was issued an Android phone running the RoadRunner application.
With Singapore’s Electronic Road Pricing system, every car owner is issued a transponder that is installed in his or her vehicle and associated with a debit account linked to a credit card. When a car enters a tollway, such as an arterial road or expressway, it passes under a reader mounted on an overhead gantry. This reader collects the transponder’s unique identifier and then deducts the appropriate amount from the driver’s account. This system is more dynamic than those used for convention road tolls because the rate a driver pays varies according to the time of day (it changes up to every half hour during peak driving times).
The fee system is primarily designed as a disincentive to drive on roads that are in high demand during the busiest periods of the day. This approach, called congestion pricing, is an increasingly popular tool to address the environmental, health and productivity problems associated with vehicular congestion. Most of these systems use RFID technology similar to that deployed in Singapore.
But Gao and Peh think there are two problems with this approach to congestion pricing. One difficulty is that it relies on erecting a costly, static reader infrastructure. The second is that although the prices change throughout the day, they are based on expected, rather than actual, demand. In other words, because a given stretch of road has historically been congested from 5:30 PM to 6:30 PM on weekdays, the fee for driving on that road might be its highest at that time, irrespective of the actual demand on any given day.
RoadRunner addresses both hurdles by relying on radios that would be embedded into drivers’ smartphones—and eventually directly into the cars—and communicate with each other, forming ad hoc networks that would limit the number of cars entering a specific stretch of highway at any given time. They do this by directing the drivers away from roadways that are already full, using turn-by-turn directions given audibly.
Rather than charging a toll as a driver enters a congestion-controlled road, the RoadRunner system uses tokens that allow only a set number of vehicles onto the road at any one time. The system works like this: When a car is nearing the congestion-controlled roadway, its radio pings the RoadRunner software over a cellular network and requests an available token. If one is available, RoadRunner assigns that token to the car. If no tokens are available, the software routes the car away from the congestion-controlled road. If the driver disregards these directions and continues onto that road, RoadRunner, using the cellular connection, issues the driver a penality—which would be a fine higher than the cost of a token, thus acting as a disincentive to ignore routing directions again in the future.
Gao and Peh’s approach would also address another form of congestion: broadband cellular traffic. “There is not a lot of bandwidth left,” Gao explains. “If you brought millions of [drivers] onto the network, it would overload it” due to the intensive computation required to constantly monitor how many cars have entered or exited a monitored roadway. Rather than having to eat up broadband by having each car constantly send its GPS coordinates to a cloud-based server via a cellular connection, however, the program has the vehicles use the 802.11p communication standard to transmit their location and unique identifier to each other over an ad hoc network. The IEEE developed 802.11p to enable vehicle-to-vehicle or vehicle-to-infrastructure communication. The latter would be used for such things as electronic tolling or safety systems wherein cars can be linked to streetlights or other infrastructure. The protocol enables faster data transmissions, over a longer range (around 6 megabits per second and a range of up 1,000 feet) compared with conventional Wi-Fi.
RoadRunner uses a cellular connection only to manage the issuance of available tokens, or to issue a fine to a driver who disregards instructions to avoid a congestion-controlled zone.
Following software simulations, the team tested RoadRunner in a live scenario in Cambridge. Each driver’s phone was linked, via a USB cable, to a Cohda Wireless 802.11p radio unit. These radios allowed RoadRunner to monitor the location and number of cars inside the region that had been designated for congestion control. Once the software determined that the maximum number of tokens had been issued, it would direct the next driver headed toward that area to turn away from it, just as it had done in the software simulation.
Inside each vehicle, they also placed a second smartphone that communicated with a remote RoadRunner server via an LTE cellular connection. The researchers did this in order to compare how long it took the two systems to do things such as issue tokens or send a driver re-routing directions. For the most part, the phones that communicated through the vehicle-to-vehicle protocol (over 802.11p) performed comparably to those communicating directly to the server.
Future Scenarios
Using the software simulations, the team showed that cars could drive an average of 8 percent faster during periods of peak traffic under the RoadRunner system for distributing traffic on city streets, compared with the congestion pricing scheme, under which cars are not prohibited from entering congested areas but are charged fees for doing so.
The test performed with actual drivers in Cambridge was far too small to impact street congestion, Gao says, but it proves that the RoadRunner software functioned as expected.
The 802.11p radio units they tested, however, are roughly the size of a hardcover book, Gao reports. The next goal, he says, is to focus on the hardware side of the RoadRunner system by significantly shrinking the radio, such that it could be small enough to embed into a smartphone.
Component supplier Delphi Automotive recently announced that General Motors just integrated its 802.11p chipset, made through a collaboration between Cohda Wireless and NXP Semiconductors, in its 2017 Cadillac models. The chipset will be controlled by a microprocessor also integrated into the car, and will be used for vehicle-to-vehicle or vehicle-to-cloud applications.
To Charge or Not To Charge
Gao and Peh designed RoadRunner as a possible alternative to a congestion-pricing scheme such as the one Singapore currently uses. By routing cars away from areas where congestion is likely to halt the smooth flow of traffic, they reason, a city would not need to charge drivers for entering controlled areas—unless a driver were to disregard the directive to turn away from that area and thereby commit a infraction.
However, RoadRunner could be deployed such that in order to be issued a token for entry into a controlled area, a driver would need to pay a fee (or have a fee deducted from his or her account).
Urban planners might find this option more appealing than deploying RoadRunner without fees. After all, congestion pricing is seen as a powerful disincentive to driving private vehicles, as well as an incentive for utilizing public transit. This shift can produce environmental and productivity benefits, but RoadRunner was not designed with these goals in mind.
“RoadRunner creates a disincentive to driving on certain roads [at certain times], but it is not a disincentive to driving, in general,” Gao says. He and his colleagues have not studied what impact RoadRunner, if used across an entire city in lieu of congestion pricing, would have on carbon emissions.
Ultimately, however, people might not even be making their own decisions about what route to take through cities. One day, the RoadRunner application (or others like it) could be used to link autonomous cars with potential riders, via their smartphones. Under this scenario, a person could stand on a street corner and hail a car using the app and keying in a destination. A car would then arrive to pick up that individual, communicating with other vehicles to determine the best route. No driver—or car ownership—would be necessary.