Eyes in Sky for BHL

PPP-1cam-N.jpg (15190 bytes)

Eyes in the Sky for Berkeley Highway Laboratory

Dan Lyddy, Benjamin Coifman, Joy Dahlgren, Joe Palen

The Berkeley Highway Laboratory (BHL) covers 2.7 miles of freeway with 4-5 lanes in each direction, including a westbound (compass southbound) carpool lane. The westbound lanes, where the eastern approach to the Bay Bridge meets the interchange of interstates 80, 580, 880, and 980 (the MacArthur Maze), are particularly interesting to researchers. Because westbound drivers often make weaving maneuvers or sudden lane-changes while trying position themselves advantageously at the entrance to the maze, on any given day the prevailing speed can drop below 20 mph for more than eight of the fourteen hours between 5 A.M. and 7 P.M.

The BHL project has the following goals: (1) to improve basic traffic sensor technologies, (2) to obtain rich, high-quality, real-world traffic surveillance data by combining data from different sensors, and (3) to develop, improve, and validate traffic simulation and prediction models using this real world data. Currently, the BHL uses two main sensor technologies: the recently installed machine vision, and magnetic loop detection.

The loop detection component consists of eight paired "speed trap" detector stations in each travel lane. These detectors measure vehicle speed, count, and occupancy directly. We use these basic measurements to estimate vehicle lengths in an effort to reidentify vehicles from station to station. We currently use "on-off" loop cards that are only capable of detecting the presence or absence of vehicles. In the future, we will upgrade these cards to generate "loop signatures," magnetic profiles of each vehicle. These signatures can be used to accurately classify vehicles or reidentify them. In the future, we intend to add other technologies, including laser sensors and instrumented probe vehicles.

For more information, and for current freeway data, see:

http://www.cs.berkeley.edu/~zephyr/freeway/

and

http://www.its.berkeley.edu/projects/freewaydata/

To download the most recent Berkeley Highway Lab research report (Benjamin Coifman and Michael Cassidy, Automated Travel Time Measurement Using Vehicle Lengths from Loop Detector Speed Traps) go to:

http://www.path.berkeley.edu/PATH/Publications/PATH/PRR-2000-12.pdf