Dynamical systems approach for queue and delay estimation at signalized intersections under mixed traffic conditions

Dynamic prediction of queues and delays at signalized intersections utilize data obtained from automated traffic sensors as inputs for Intelligent Transportation Systems (ITS) applications. Errors are inevitable during automated data handling, especially when the traffic involved is heterogeneous and lacking in lane discipline. This paper presents a model-based estimation scheme that can handle both mixed traffic conditions and erroneous detector input data to estimate queue and, in turn, delay. The models were developed for Queue within Advance Detector (QWAD) and Queue beyond Advance Detector (QBAD) scenario. The statistical properties of detector errors were incorporated into the estimation scheme, and the scheme was tested for varying traffic conditions. The estimation scheme’s performance was evaluated using field data from a signalized intersection in Chennai, India, and simulated data. It is found that the incorporation of statistical properties of detector error allows accurate estimation of queues and delays despite erroneous data input.