Application of a Dynamic Model for Arterial Street Incident Detection

Abstract

Existing freeway and signalized arterial street incident detection algorithms were investigated to determine their merit for use on urban arterial streets. Based on this investigation, a Kalman filtering algorithm was modified to recursively filter and update aggregate traffic flow and speed data to estimate true values. A test using measured arterial street data at a signalized intersection shows good tracking ability on these traffic variables over time. A test using data from an incident on an arterial street also confirmed that this algorithm has good potential for arterial street incident detection.