Citation

Furth PG, Halawani ATM. Transp. Res. Rec. 2016; 2586: 83-89.

Copyright

(Copyright © 2016, Transportation Research Board, National Research Council, National Academy of Sciences USA, Publisher SAGE Publishing)

DOI

10.3141/2586-09

PMID

unavailable

Abstract

If the vehicular phase concurrent with a pedestrian phase is running under fully actuated control or is a noncoordinated phase under coordinated-actuated control, the length of its green is not known when the phase begins; it could run anywhere between a (known) minimum and maximum green, depending on when a gap is detected. In such a case, walk intervals are generally kept to a minimum length to prevent the pedestrian phase from constraining the ending of green that can lead to situations in which the pedestrian phase has cleared yet the concurrent phase continues its green for 20 more seconds. Proposed is the concept of an adaptive walk interval whose length is based on the predicted length of the concurrent phase's green, which in turn is calculated cycle by cycle with data from recently past cycles. That way, during heavy traffic periods in which the vehicular phase usually goes well beyond its minimum, a longer-than-minimum walk interval can be provided with very little impact on signal operation or vehicular delay. Two methods, ratio estimation and stratification, are proposed and tested for predicting the green time needed by a vehicular phase; both methods use as data only traffic signal timing data from the last several cycles. Simulation tests with coordinated-actuated and fully actuated control, with and without pedestrian recall, and with and without permissive windows show that adaptive walk intervals can markedly reduce pedestrian delay with almost no impact on vehicular delay.

Language: en