Tomer Toledo and Haneen Farah introduce a new project designed to enable a better understanding of factors influencing the decision whether or not to overtake
Two-lane highways make up a substantial proportion of the road network in most of the world. About 60 per cent of all fatal crashes in OECD member countries occur on such roads (OECD, 1999).
Passing is among the most significant driving behaviours on two-lane highways. It is a mentally complicated task that substantially impacts on highway performance. A reduction in passing opportunities leads to the formation of vehicle platoons, which in turn cause a decrease in the level of service and negatively affect safety, fuel consumption and air quality. Potential improvements to the design of two-lane highways include construction of additional lanes, passing sections, 2+1 lane designs and widening of existing lanes and shoulders. However, these solutions are costly and require careful design and evaluation prior to implementation. Thus, a better understanding of passing behaviour is essential.
In order to collect data on drivers’ passing behaviour, a laboratory experiment using a driving simulator was developed. STISIM, a fixed-base interactive driving simulator, was used in this experiment, which focused on the subject’s decision whether or not to pass the vehicle in front. In making this choice, the subject needs to consider the available passing gaps. These gaps are defined as the time gap between the opposing vehicle and the vehicle in front of it, at the time this vehicle encounters the subject vehicle.
Collecting behaviour data
In order to capture the impact of various infrastructure and traffic factors on passing behaviour, a number of different simulator scenarios were designed. All scenarios in the experiment included 7.5 km two-lane highway sections with no intersections. The sections were on level terrain and in daytime and good weather conditions, which allowed for good visibility. Drivers were instructed to drive as they would normally do in the real world and were first given 5 to 10 minutes to become familiar with the simulator.
100 drivers (69 males, 31 females) who had held a driving license for at least five years and drove on a regular basis participated in the experiment. The age of the participants ranged between 21 and 61 years. The simulator collected data on the longitudinal and lateral positions, speeds and acceleration of the subject vehicle and all other vehicles in the scenario at a resolution of 0.1 seconds. From this raw data, other variables of interest were calculated. The resulting data set included a total of 14,654 passing gap observations. In 696 (4.7 per cent) of these gaps, the drivers completed passing manoeuvres.
The completion of passing manoeuvres is modelled in two stages. Drivers are first assumed to decide whether or not they want to pass the lead vehicle. Drivers that are interested in passing evaluate the available passing gap and either accept it and pass the vehicle in front, or reject it and refrain from passing.
It was found that the desire to pass is affected by the difference between the subject’s desired speed and the current speed of the vehicle in front. This difference captures the extent to which the front vehicle imposes a constraint on the speed of the subject. In the data, the desired speed for each driver was calculated as the mean speed of the vehicle in the sections during which it was not close to the vehicle in front. It was also found that the desire to pass was higher when the distance between the subject and the front vehicle was reduced.
The collection of driving simulator data may lead to biases in behaviour. For example, simulator drivers may be indifferent or become tired with the experiment as it progresses and so modify their behaviour. A cumulative distance variable, which is defined as the total distance the subject has driven from the beginning of the experiment to the measurement point, was introduced in the model in order to correct this effect. It has a small, but significant, positive effect on desire to pass probability. Thus, the desire to pass increases as the experiment progresses, possibly in order to enable the subject to complete the task sooner.
Taking the leap
Passing gap acceptance decisions are affected most by the variables related to the speeds of the subject vehicle, the vehicle in front and the opposing vehicle. Mean critical passing gaps are larger when the speed of the subject vehicle is lower and when the speed of the vehicle in front is higher. This is intuitive because it is more difficult to complete the passing maneuver, as it requires more time and longer distances.
The type of vehicle in front also affects the critical gaps. It is larger for trucks, which obscure the field of vision and pose a higher safety risk, compared to passenger cars. In contrast, critical gaps decrease when the speed of the opposing vehicle increases. Note that critical gaps are measured in time units. Therefore, higher speed of the opposing vehicle results in larger critical gaps in terms of distance. Figures 1 and 2 illustrate the impact of the speeds of the subject vehicle and the vehicle in front on the mean critical gap, both in terms of time and distance. In each of these figures, one variable was varied while all the other variables were fixed. The critical gaps were calculated for a female driver over 25 years of age driving on a tangent section. Unless varied, the figures assume that the subject’s speed is 80 km/hr (22.2 m/sec); the speed of the vehicle in front is 60 km/hr (16.7 m/sec); and that the speed of the opposing vehicle is 90 km/hr (25.0 m/sec).
The geometric design of the road also affects passing behaviour. In this model, this is captured by the road curvature. Critical gaps are smaller in roads with large curve radii, which allow larger sight distances, compared to roads with tighter curves. Critical passing gaps vary substantially in accordance with driver characteristics. They are significantly smaller for younger drivers compared to older ones. The gender of drivers was not found to be statistically significant. Finally, the size of the available gap clearly affects passing gap acceptance, with higher acceptance probabilities for larger gaps.
Latent driver characteristics
An individual-specific error term was introduced in the model to capture latent driver characteristics. It was statistically significant in both parts of the model. The parameters of this term were positive in the desire to pass model and negative in the gap acceptance model. This result is consistent with an interpretation of the term as representing aggressiveness and level of skill. Aggressive drivers are more likely to desire to pass, and accept lower critical gaps, compared to timid drivers.
While the results reported here are promising, this work has limitations that merit further research in several directions. Possibly the most important limitation is that the model estimation used only data from a driving simulator. The estimation results need to be validated against real-world data to eliminate biases resulting from the use of the simulator. The affect of the geometric design of the road on passing behaviour was captured only through the road curvature. This is partly because important design parameters, such as those related to the quality of the pavement, sight distances or roadside features, are difficult to model and also to perceive in the simulator.
In addition to improving our understanding of drivers’ behaviour, the intended practical application of the model presented is ultimately in the framework of traffic simulation models. This would require additional extensions to handle situations such as aborted passing manoeuvres and overtaking multiple vehicles in a single pass. Finally, safety indicators related to passing manoeuvres need to be developed, and the impacts of different geometric, traffic and driver characteristics on the risks and severity of car crashes need to be further studied.
About the authors:
Tomer Toledo is senior lecturer at the Technion Faculty of Civil and Environmental Engineering - Israel Institute of Technology, Haifa, Israel. Haneen Farah is Dr. of Transport Safety at the Department of Transport and Economics, Royal Institute of Technology (KTH), Stockholm, Sweden.