Variable message signs (VMSs), as one of the important ITS devices, provide real-time traffic information of road network to drivers in order to improve route choice and relieve the traffic congestion. In this study, the effectiveness of VMS on driving behavior was tested based on a driving simulation experiment. A road network with three levels of VMS location to route-diverging intersection and three types of VMS information format was designed in a high fidelity driving simulator platform. Fifty-two subjects who were classified by driver age, gender, and vocation successfully completed this experiment. The experimental results showed that driver characteristics, VMS location, and information format profoundly influence driving behaviors. Based on the research findings, it is suggested that VMS would be positioned between 150 m and 200 m upstream of the diverging point to balance the VMS effects on traffic safety and operation and the graphic information VMS format is better than the format with text massage only.
With the rapid development of urbanization and motorization, China has become the world’s second largest car country and the constantly climbing number of vehicles in urban road networks leads to more and more serious traffic congestion problems. The traditional countermeasures of traffic congestion alleviation include constructing new roadways, adding new traffic facilities, and strengthening traffic management [
Variable message signs, as an advanced traffic guidance system, can provide real-time traffic information in urban road networks to help drivers choose the routes with lower traffic volumes. Thus, the vehicles can be distributed reasonably in road networks so as to improve the performance of traffic system [
A number of previous studies focusing on VMSs and relevant driving behaviors have been conducted. Two typical research methods, questionnaire, and computer simulation experiment have been applied for analyzing main factors that influence drivers’ route choices in the VMS environments. Through the questionnaire, it was found that whether drivers accept an item of VMS advice or not is closely associated with drivers’ characteristics and their familiarity degree to the road network [
While the statement preference (SP) methods of questionnaire and computer simulation experiment are limited to provide detailed driving behavior data, driving simulators were applied to investigate drivers’ speed control, lane change, and response to VMSs in virtual reality road environments. Driving simulators can provide a well-controlled experimental condition to compare the drivers’ behaviors in road networks with different VMS settings. Another advantage of using driving simulator is that it can collect the data which are difficult to achieve in the real world, especially the vehicle’s instantaneous velocity [
Although a few of previous studies involved drivers’ behaviors in the VMS traffic environment, there is a lack of research focusing on investigating how VMS position and information format affect driving behavior, especially using a high fidelity driving simulator. The main objective of this paper is to investigate whether and how VMS position and VMS information format impact drivers’ behaviors, such as route choice, speed control, and lane changing, based on a high fidelity driving simulation experiment.
A total of 57 test subjects were recruited in this experiment. Every subject would encounter the VMSs six times and there were 342 samples that reflect the drivers’ performances under VMS. The similar scale of sample sizes was also applied in the driving simulator experiments by previous simulator experiment designs [
Gender and vocation distribution of recruited subjects.
Gender | Professional | Unprofessional | Total |
---|---|---|---|
Male | 16 | 14 | 30 (57.7%) |
Female | 9 | 13 | 22 (42.3%) |
|
|||
Total | 25 (48.1%) | 27 (51.9%) | 52 (100%) |
A high fidelity driving simulator located in MOE Key Laboratory for Urban Transportation Complex Systems of the Beijing Jiaotong University was used in the study, as shown in Figure
The driving simulator.
In addition, five cameras are installed inside and outside the cabin to supervise the experimental process. An emergency stop button is installed both inside cabin beside the driver seat and in the front of control desk in order that either subject or researcher can discontinue the experiment immediately in case the subject suffers driving simulation sickness.
To investigate the effectiveness of VMS on driving behavior, a road network was designed, as shown in Figure
Scenario | The first part | The second part |
---|---|---|
VMS-I location | VMS-II information format | |
A | 0 m | Text-only format |
B | 200 m | Graphics-only format |
C | 400 m | Combination of text and graphics |
The experimental road network.
In each driving scenario, subjects needed to drive from the start point to the end points, as illustrated in Figure
The design of traffic congestion and driving scenario.
Upon arrival, the subjects were asked to fill out and sign an informed consent form (per IRB). The subjects were then advised to drive and behave as they normally would and to adhere to traffic laws as in real life situations. The subjects were also notified that they could quit the experiment at any time in case of driving simulation sickness or any kind of discomfort. Prior to the formal experiment, drivers were trained for at least 10 min to familiarize with the driving simulator operation and the experimental road network. During the course of the practice, subjects exercised selected maneuvers including straight driving, acceleration, deceleration, left/right turn, and other basic driving behaviors. Then, the formal experiments began during which all subjects would test the three scenarios A, B, and C in a random order so as to eliminate the experimental time order effect. For security and liability reasons, each subject was escorted to the simulator cabin to commence the experiment and he/she was allowed at least 20 min to rest before running the next scenario.
Data collection and analyses were based on each subject driving three times in the simulated road network. Each subject would meet the different types of VMSs six times for a total of 312 route choices. The related dependent measures for driving behavior analyses were defined as follows: RC (straight = 0; turn = 1): route choice, whether a driver went straight or turned at an intersection; SPEED (km/h): the vehicle’s average speed for every twenty meters upstream or downstream of the VMS; SV (km/h): speed under VMS, the vehicle’s operation speed under VMS; LCT (s): lane changing time, which was measured as lane changing duration if a subject has a lane changing behavior; LCP (m): lane changing position, which was measured as the distance to the intersection at which a subject started changing lane if a subject has a lane changing behavior; LCL (m): lane changing length, which was measured as the longitudinal distance of lane changing; LCS (km/h): lane changing speed, which was measured as the average speed during lane changing; LCD (m/s2): lane changing deceleration, which was measured as the average deceleration during lane changing.
Based on the driving behavior data, the following results focused on studying the effects of VMS position, VMS information delivery formats, subject vocation, gender, and age on driving behaviors. The hypothesis testing in the following analyses are based on a 0.05 significance level.
The binary logistic model is suitable for analyzing route choice behavior because the behavior can be described as a dichotomy variable. The binary logistic regression technique has been applied to explore the relationship between route choice and its potential influencing factors [
Parameter estimates of logistic regression models for route choice.
Mode | Variable | Level |
|
S.E. | Wald | Df | Sig. | Exp( |
---|---|---|---|---|---|---|---|---|
The first part (VMS-I) | Age | Continuous | −0.079 | 0.020 | 16.212 | 1 | 0.000 | 0.924 |
Gender | Male versus female | 0.968 | 0.370 | 6.836 | 1 | 0.009 | 2.633 | |
Location | — | — | — | 4.347 | 2 | 0.114 | — | |
400 m versus 0 m | −0.873 | 0.439 | 3.953 | 1 | 0.047 | 0.418 | ||
200 m versus 0 m | −0.200 | 0.430 | 0.216 | 1 | 0.642 | 0.819 | ||
SV | Continuous | −0.033 | 0.013 | 6.065 | 1 | 0.014 | 0.968 | |
Constant | 4.682 | 1.341 | 12.197 | 1 | 0.000 | 107.99 | ||
|
||||||||
The second part (VMS-II) | Age | Continuous | −0.112 | 0.042 | 7.272 | 1 | 0.007 | 0.894 |
Vocation | Yes versus no | 2.066 | 0.820 | 6.351 | 1 | 0.012 | 7.896 | |
Format | — | — | — | 4.132 | 2 | 0.127 | — | |
Text-only versus Text and graphic | −0.885 | 0.442 | 3.998 | 1 | 0.046 | 0.413 | ||
Graphic versus Text and graphic | −0.279 | 0.425 | 0.431 | 1 | 0.511 | 0.756 | ||
SV | Continuous | −0.057 | 0.016 | 12.147 | 1 | 0.000 | 0.944 | |
Constant | 6.597 | 1.775 | 13.816 | 1 | 0.000 | 733.140 |
In the first part for VMS-I analysis, the regression results indicate that the independent variables of age (
In the second part for VMS-II analysis, the regression results indicate that the independent variables of age (
According to the two logistic regression models, VMS location, VMS information formats, and speed under VMS are other three important factors that influence drivers’ route choice decision. In order to illustrate how VMS location and VMS information format impact the route choice behavior, Figure
Probability of turn decision based on the logistic regression models.
VMS location
Information format
In terms of speed control behavior, this analysis focuses on the driving speed around VMS-I. The mean of the speed under VMS is 67.59 km/h, and the standard deviation is 13.12. In the cases that VMS-I’s location is 0 m and 200 m from the intersection, the drivers’ speed control behavior would be influenced by the intersection because they often decelerate to negotiate with the slow downstream traffic. This may confuse the VMS’s effect on speed control. Therefore, the scenario C (VMS-I’s location is 400 m from the intersection) is used for the analysis of speed behavior in the VMS environment. The average speed curve before and after VMS-I is shown in Figure
The trend of average velocities before and after VMS-I.
The measures of LCT, LCP, LCL, LCS, and LCD are used for exploring how VMS position and information format affect drivers’ lane changing behaviors in different scenarios. The basic statistical descriptions for LCT, LCP, LCL, LCS, and LCD are summarized in Table
Descriptive statistical results for LCT, LCP, LCL, LCS, and LCD.
Variable | LCT | LCP | LCL | LCS | LCD | |||||
---|---|---|---|---|---|---|---|---|---|---|
Mean | S.D. | Mean | S.D. | Mean | S.D. | Mean | S.D. | Mean | S.D. | |
Location | ||||||||||
400 m | 12.3 | 4.0 | 246.9 | 187.8 | 151.5 | 94.5 | 12.1 | 6.9 | 0.5 | 0.4 |
200 m | 14.1 | 6.5 | 244.5 | 155.6 | 160.2 | 101.5 | 11.5 | 5.2 | 0.5 | 0.4 |
0 m | 13.2 | 5.7 | 179.0 | 65.8 | 145.2 | 65.9 | 11.2 | 3.6 | 1.1 | 0.7 |
|
||||||||||
Format | ||||||||||
Text | 10.3 | 4.2 | 68.7 | 50.6 | 62.5 | 44.9 | 6.0 | 2.8 | 0.9 | 0.6 |
Graphics | 14.3 | 5.4 | 158.5 | 99.1 | 129.7 | 78.5 | 9.6 | 5.5 | 0.7 | 0.3 |
Text and graphics | 11.7 | 5.1 | 179.2 | 99.4 | 136.2 | 67.0 | 12.0 | 4.8 | 0.9 | 0.7 |
MANOVA variance analysis of lane changing behavior.
Dependent variable | Independent variable | Type III SS | DF | Mean square |
|
Sig. |
---|---|---|---|---|---|---|
The first part: VMS location (VMS-I) | LCT | 38.32 | 2 | 19.16 | 0.625 | 0.539 |
LCP | 50245 | 2 | 25212 | 1.085 | 0.345 | |
LCL | 2264.30 | 2 | 1132.15 | 0.134 | 0.875 | |
LCS | 7.92 | 2 | 3.96 | 0.131 | 0.878 | |
LCD | 3.74 | 2 | 1.87 | 8.249 | 0.001 | |
|
||||||
The second part: VMS information formats (VMS-II) | LCT | 142.03 | 2 | 71.01 | 2.890 | 0.065 |
LCP | 119829 | 2 | 59915 | 7.733 | 0.001 | |
LCL | 57245.30 | 2 | 28622.69 | 6.622 | 0.003 | |
LCS | 322.44 | 2 | 161.22 | 7.762 | 0.001 | |
LCD | 0.38 | 2 | 0.19 | 0.602 | 0.551 |
For the VMS-I, only LCD is significantly influenced by VMS location (
The influence of the VMS location on LCD.
For the VMS-II, it is found that LCT (
The influence of the VMS information formats on LCT, LCP, LCS, and LCA.
LCT (lane changing time)
LCP (lane changing position)
LCL (lane changing length)
LCS (lane changing speed)
The main purpose of this study is to investigate how VMS position and information format affect route choice, speed control, and lane changing behaviors in the road network using a high fidelity driving simulator. Three levels of distances ranging from 0 to 400 m between VMS location and route-diverging intersection were designed and three kinds of information formats including text-only, graphics-only, and combination of text and graphics were tested in the simulation experiment. The experimental results showed that both VMS location and information format profoundly influence driving behaviors.
In the past two decades, various experiments applied driving simulators to study the impact of VMS on the route choice behavior. Based on the experimental data, some route choice models were developed to evaluate the drivers’ route choices under VMS and enhance network performance [
A prior study focused on how the content of VMS affected the driving behavior, which indicated that the VMS information content, including the level of detail of relevant information, socioeconomic characteristics, network spatial knowledge, and confidence in the displayed information, significantly affected drivers’ willingness to divert [
Additionally, the result analyses indicated that the driver characteristics of age, gender, and vocation also have significant effects on route choice behavior in the VMS environments. It was found that the older drivers are less willing to change driving route under the VMS guidance; compared to female, the male drivers are more likely to be influenced by VMS; and the professional drivers are more likely to accept the information released by VMS.
In summary, this paper explored the relationship between driving behaviors and VMS’s position and information format based on the driving simulator experiment. The findings of this study would be helpful for traffic engineers to select VMS installation locations and design VMS information delivery formats in order to optimize traffic safety and efficiency in urban road networks.
The authors declare that there is no conflict of interests regarding the publication of this paper.
This work is financially supported by “863” Research Project (2011AA110303).