Berkeley - Drivers who rely on radio traffic reports may soon have an easier way to navigate the dreaded freeway commute.
A University of California, Berkeley, professor and his team of students have developed a way to get updates on traffic hotspots, alternative routes and travel times - up to an hour in advance - via the Internet or cellular phone.
The freeway Performance Evaluation Monitoring System, or PeMS, will be unveiled tomorrow (Thursday, Jan. 18) at the UC Berkeley-sponsored Bay Area transportation town meeting. Developed by Pravin Varaiya, UC Berkeley professor of computer science and electrical engineering, PeMS converts freeway monitoring data into real-time traffic updates accessible via a Web portal.
Varaiya is one of the main researchers in the proposed UC Berkeley-led Center for Information Technology Research in the Interest of Society (CITRIS). A joint program with UC Santa Cruz and UC Merced, CITRIS is a research initiative dedicated to creating technology to improve everyday living.
Currently PeMS is only available to engineers at the California Department of Transportation (Caltrans) and to the research community. But Varaiya hopes his invention will soon enjoy widespread use through Internet service providers and public agency traveler information efforts.
At the heart of Varaiya's invention is software that converts data from Caltrans' existing vehicle detection network into easy-to-read tables and graphs. The PeMS Web page provides a map of the entire freeway system in a given urban area. A color-coded link provides the freeway speed, and an animation shows how congestion starts and spreads. By selecting an origin and destination on the map, the user can see how long each route to the destination will take. PeMS also analyses traffic patterns and predicts travel times up to an hour in advance.
While PeMS has obvious advantages for commuters, Varaiya, who is UC Berkeley's Nortel Networks Distinguished Professor, originally designed the system to help Caltrans officials monitor traffic patterns. "PeMS can be used by managers to get an overall view of traffic trends, by engineers to spot and eliminate bottlenecks and help them design solutions, by planners to evaluate proposals for new routes," Varaiya said.
Caltrans officials have already begun using PeMS. "The real benefit of PeMS is it gives us a better understanding of what is happening on the freeways," said John Wolf, chief of the Office of System Management Planning in the Caltrans Traffic Operations Program.
According to Varaiya's calculations, appropriate traffic management in the Los Angeles area could save more than $1 billion per year in lost time and fuel costs. And this estimate did not include costs associated with pollution, increased accidents or stress.
Using PeMS, Varaiya found that the most efficient way to reduce commute times is to keep freeway traffic moving at 60 miles per hour. The best way to maintain this speed during peak use hours, he said, is to limit the number of cars allowed onto the freeway. This could mean longer waits at on-ramps, but, overall, the trip time would be shortened.
Varaiya envisions a system where drivers could be notified of the best time to leave for the journey. "Instead of being parked on the freeway, you could spend ten more minutes on your coffee break or in your office," he said.
In addition to its use in traffic management, PeMS can predict travel times. For example, a 40-mile trip across Los Angeles on Interstate 10 can take from 40 to 130 minutes, depending on traffic. By accessing PeMS, a traveler could inform business clients of her arrival time.
Although the software for each urban area must be customized, a new Caltrans district can be added within weeks. Varaiya and his students are currently adapting software for Los Angeles, which generates one gigabyte (one billion bytes) of data per day. The entire state generates an average of two gigabytes per day.
PeMS is not yet applicable in the Bay Area, where traffic has worsened dramatically in the past five years, said Varaiya. But Caltrans has recently increased availability of data from the detectors at least 10-fold, said Judy Chen, chief of the Caltrans Office of Traffic Systems.
Caltrans also feeds data to TravInfo, a traffic information system - dial 817-1717 in the nine-county Bay Area - maintained by the Metropolitan Transportation Commission. But PeMS provides more detailed analyses that can be used for research.
Each vehicle-counting device, or loop detector, consists of an electrical wire buried in the pavement. When a vehicle passes over the wire, the metal in the vehicle causes fluctuations in the electric current. A detector that monitors a four-lane highway can cost $100,000 or more.
For PeMS, the loop detectors send data every 30 seconds to a Sun 450 computer workstation in the basement of UC Berkeley's Cory Hall. PeMS software combines the 30-second readings into 5-minute readings that yield the number of cars and their average speed for each lane, and then combines the lane-by-lane data into an overall measure of freeway flow, occupancy and speed.
Traffic monitoring systems like PeMS can accommodate future demand for the next five years, Varaiya said. However, long-term solutions will require automated vehicle guidance systems like those being developed by California Partners for Advanced Transit and Highways (PATH), a collaborative public and private effort to apply advanced technology to reducing traffic congestion, air pollution and energy consumption.
Meanwhile, Varaiya is working with UC Berkeley computer science professor Jitendra Malik to add video monitoring to PeMS. This project involves placing video monitors atop buildings to track vehicle speed and direction. Eventually, through CITRIS, the proposed UC Berkeley center, Varaiya and Malik plan to expand their traffic monitoring research to deploy tiny wireless sensors that can beam traffic data to a central computing facility, yielding even more precise estimations of traffic flow.
PeMS is supported by Caltrans and the National Science Foundation through California PATH.