The Fence

Queue length estimation at signalised intersections

Queue length estimation

This PhD project proposes a methodology to estimate real-time queue lengths by using high-resolution detector data and signal timing data. Shockwave theory is considered as a building block to identify traffic state changing points (e.g., capacity, jam density, free flow) and to calculate the maximum queue length at each signal cycle.

Real-time queue length is crucial information, increasingly needed for signal operation and signal optimisation purposes. Currently two distinct types of models are used to tackle the real-time estimation of vehicle queues in signalised intersections. The first is based on the analysis of cumulative traffic input–output to a signal link and the second type of models are constructed based on the analysis and modelling of traffic shockwaves.

The first approach has significant drawbacks in estimating long queues exceeding beyond a vehicle detector. The second approach builds on shockwave theory and can successfully explore the complex queueing process in both temporal and spatial dimensions assuming a known vehicle inflow.

An extension to the proposed methodology is developed within the Kalman filter framework to minimise the uncertainty caused by detector measurements, thus enhancing the model reliability. This approach can estimate time-varying queue length even when the signal links are over-saturated.


See the full list of iMOVE projects here

Project background

Queue length is one of the most crucial performance measures for signal control evaluation or signal optimisation. Over the years, many researchers have devoted themselves to the study of queue length, which can be divided into three categories (i.e., detection, estimation, and prediction), according to queue length acquisition methods.

The first category – the direct detection of queue length using equipment such as cameras – is one of the commonly used methods to obtain queue length in recent research. This method can simply and quickly obtain the queue length, but it does not consider fluctuations in traffic flow, and the maximum queue will not be observed when the queue length exceeds the visual range of the camera.

The second category, queue length estimation, is the most studied by scholars. In the literature, queue length estimation methods can generally be classified into two categories: input-output models and shockwave models. The input-output models analyse the cumulative traffic input-output (arrival-departure curve) of a link to estimate the queue length. This model has simple conceptual properties. It is, however, limitedby the inability to capture the spatial queue in actual arterial traffic.

Additionally, the traditional input-output analysis cannot describe the spatial distribution of queue length in real time, nor is this model suitable for the estimation of queue length at oversaturated intersections. Recently, much attention has been given to the formation and dissipation of queues using traffic shockwave theory. The shockwave model provides a better analysis framework for queue length estimation.

With the advances in data collection technologies, the estimation of queue length by probe vehicles has also become a common method. Because of the unique mobility of probe vehicle data and limitations on probe vehicle size, the precision of queue length estimation can be guaranteed only when the penetration rate of probe vehicles is high. Recently, estimating real time arterial performance measures using more fine-grained signal timing and detector data has gained much attention.

The third category is the prediction of queue length. With the improvement of traffic control requirements, predictive traffic signal control has become a developing trend, which relies on the ability to obtain relevant parameters of traffic control in advance. Therefore, the prediction of queue length is essential for predictive control optimisation.

Recently, predicting and estimating real time arterial performance measures using more fine-grained signal timing and vehicle loop detector data has gained much attention. The selection of an appropriate methodology to estimate or predict the queue lengths at signalised intersections depends on the type of data sources available and its resolution for real world applications.

Most frequently available data type in real world are flow, occupancy and speed data obtained via inductive loop detectors placed on road segments with different resolutions. Further, with the recent technological advancements, other types of data sources such as vehicle actuation data, probe vehicle data, GPS trajectory data, Bluetooth and radio frequency data are used when available. The Brisbane Road network (both motorways and arterial roads) is equipped with vehicle loop detectors, mostly placed close to the stop line of each intersection.

These vehicle loop detectors provide flow and occupancy data every 2 seconds. Such high-resolution data related to traffic parameters of a particular road segment can be used to identify the traffic state changing points (such as from arrival state to saturation state and to capacity state) occurring with the effect of stop and go nature happening at an intersection.

Firstly, this research will focus on developing a methodology to estimate and predict the cycle-based maximum queue lengths by only using the high resolution (2-seconds occupancy and flow) data in a framework of identifying the traffic state change points coupled with the shockwave theory. How to account for errors and uncertainties in the identified traffic state changing points will be the primary focus under the first objective, and a Kalman filter-based methodology is proposed to minimise the error and update the estimates in a recursive manner.

Secondly, as an extension to the first objective, the possibility of making the model more robust by using other types of data sources (mainly the travel time and location details collected via Bluetooth detectors and probe vehicle data) will be investigated.

Project objectives

Two objectives are formulated to bridge the research gap:

  1. Develop a method to estimate the signal cycle-based maximum queue length at signalised intersections by using limited data sources.
  2. Develop a robust queue length estimation and prediction method applicable to isolated signalised intersections by data fusion.

More from iMOVE

Most Popular

Newsletter Signup

The Fence
To Top