Advanced Traffic Signal Control Algorithm

Although a majority of the large cities in the United States employ pre-timed traffic control, it is estimated that almost 80 percent of all traffic signals in the nation use traffic detectors, mainly for intersection control but with a small number for traffic responsive system operations. Regardless of the principle of operation, these systems essentially rely on "point" detection and therefore provide binary information on the presence or absence of vehicles. The traffic signal control algorithms based on this type of vehicle information were developed in the late 1970s and are, with minor modifications, still in operation today. There is a significant gap between the theory - the well-developed control methodologies - and real-world practice in traffic control. This gap comprises our technical rationale and elements of the gap include the fact that: (1) Existing models inadequately describe real-world traffic streams in a signalized road network, especially at near-capacity or oversaturated conditions; (2) computational complexity for optimization is too high to apply to a fair-size network, and (3) control actions have until now are based on flow and occupancy observations at fixed and limited locations. Several forms of the fundamental diagram of traffic flow (traffic flow vs. density relationship) have been proposed and experimentally verified for freeways to describe the system state and be used for traffic control (typically, ramp metering). However, the understanding of traffic flow dynamics on networks controlled by traffic signals is still limited, which makes the development of system-wide control strategies difficult to develop and implement. The signalized arterial traffic network is a complicated, nonlinear, high-order dynamic system. Its current state is defined by the locations and velocities (speed and heading angle) of all of its vehicles and the phases of all of its signals. Moreover, for control purposes it is also valuable to have preview knowledge of impending changes in vehicle movements (e.g., turning, stopping at mid-block destinations) and signal phases. Current traffic signal control systems have been constrained to make decisions based on extremely limited knowledge of the state of the network, essentially by detection of the presence of vehicles at a few discrete locations (within 10 m of the stop bar and, for some algorithms, at one mid-block location as well).


    • English


    • Status: Active
    • Funding: $999871.00
    • Sponsor Organizations:

      California Department of Transportation

      1227 O Street
      Sacramento, CA  United States  95843
    • Project Managers:

      Siddiqui, Asfand

    • Performing Organizations:

      Regents of the University of California, Berkeley

      Berkeley, CA  United States 
    • Principal Investigators:

      Skabardonis, Alexander

    • Start Date: 20100401
    • Expected Completion Date: 0
    • Actual Completion Date: 20120630
    • Source Data: RiP Project 27553

    Subject/Index Terms

    Filing Info

    • Accession Number: 01466100
    • Record Type: Research project
    • Source Agency: California Department of Transportation
    • Files: RiP, STATEDOT
    • Created Date: Jan 3 2013 3:10PM