Driver behaviour: the key to analyze traffic functionality

Traffic assessment studies the functionality of traffic infrastructures, e.g. roads, traffic signalised intersections or intersections with restrictions, etc. This type of analysis focuses on the interactions between traffic participants (motorised/non-motorised) and infrastructure in order to reveal the relationship between traffic participants and the resulting functionality.

For this reason, traffic assessments are empirical studies that rely heavily on high quality measurements of real data, including the following aspects:

– Where: the geometric aspect, e.g. direction, lane,…

– How much: traffic density and intensity.

– Who: users of the infrastructure, type of vehicle, pedestrians, etc…

– When: periods, e.g. rush hour vs. off-peak hour, or holiday vs. working day…

– How: driving behaviour parameters, e.g. speed, headway,…

– What: under specified condition, e.g. rain, accident,…

Among them, the driving behaviour factors referring to speed, gaps, when to merge, or how to act on the studied infrastructure are the most complex data to obtain from the real world.

Not only because they are difficult to collect under different environments without disrupting the traffic on site, but also because they are very complex to process into raw data. Because of this, such behavioural data is often overlooked or defaults are used in traffic analysis.

Many new field data collection technologies and devices have been introduced in the field such as loop detectors, and devices based on radar, GPS, or video cameras, including UAV (Unmanned Aerial Vehicle) devices. Collecting data on driving behaviour becomes more feasible and accurate.

Among the mentioned devices, UAV (known as drone) is more and more used for traffic flow and driving behaviour as it does not need additional equipment on the ground and, being non-invasive, does not influence the driver’s behaviour. It can capture drivers’ tactical manoeuvring decisions in different traffic conditions, such as free flow/congested, normal/rainy conditions, commuting/non-commuting conditions, due to its cost and data quality, with reasonable cost.

As Vectio is so familiar with data collection using UAV, we have a great experience to deal with videos that may contain failures in the process, such as bad light, unstable or rainy conditions.

There are several programs that allow modellers to process video recorded by UAVs. Modellers could obtain information not only about traffic counts for each movement and each vehicle type, but also related to driving behaviour, such as speed, headway, acceleration, lane change and gap acceptance.

One of those we have recently explored is the Goodvision software^[1]. which provides traffic modellers with an interactive platform to obtain the above mentioned information behavioural parameters through a video processing approach. Readers who are interested in learning more about Goodvision’s services can consult their website and Daniel’s blog^[2].

This blog shows an example of how we obtain the critical gap using the data collected by UAVs. The critical gap is an important parameter used to calculate the capacity and delay of a road in unsignalised intersection gap acceptance theory. It is defined as the minimum time (gap) between successive vehicles in the opposing (major) traffic stream that is acceptable for the entry of opposing (minor) stream vehicles (SIDRA User Guide). The accuracy of the capacity estimation is mainly determined by the accuracy of the critical gap.

The following 4-step method is proposed to estimate the critical gap of a roundabout, so that the required accepted gaps and maximum rejected gaps can be obtained first, and then the critical gap can be calculated using the maximum likelihood estimation technique.

– Preliminary observation of traffic density and collection of the approach/circulation speed of each arm of the roundabout to filter traffic movements with successive flows (Figure 1 and Figure 2). It needs to be observed; both the accepted gap and the maximum rejected gap can be obtained on the move with a successful traffic flow. The analyst should set the threshold as a minimum tracking margin to detect the successful traffic flow. The threshold is variable depending on the type of traffic infrastructure.

– Define the gap event. Through the platform, the analyst could configure the stop zone to collect the gaps of the minor flow vehicles and the circulation line to obtain the gaps of the main traffic flow (Figure 3).

– Develop your own method to obtain the fleet of vehicles circulating inside the roundabout, and the time (headway) of each vehicle entering or waiting to access the roundabout. In this way, it is possible to calculate the gaps accepted and the maximum gaps rejected (see graph 1).

– Calculate the critical gap for each lane and each traffic movement using the maximum likelihood method.

The maximum likelihood method (MLE) is applied to estimate the critical gap which is based on the fact that a driver’s critical gap is between the range of his largest rejected gap and his accepted gap. A probabilistic distribution for critical gaps must be assumed. The reader can find more information in the work of Tian et al.

In this way, the obtained critical gap can be used to calibrate and validate microsimulation software such as Aimsun, VISSIM, and thus cover the whole range of traffic on the same road configuration. Finally, all the collected data can be adopted to evaluate the traffic operation according to the level of service.

In addition to the methodology presented in this blog that uses a drone video and the Goodvision platform to extract driving behaviour parameters, there are also other resources with data from publicly available trajectory datasets, e.g. the highD dataset (German), the NGSIM dataset (USA) and open data portals (Canada), can contribute significantly to this period of prosperity of traffic flow studies.ç

Yang Wang | PhD. Transport Planner