Evaluating Driver Response to an Advanced Speed Display near Uncontrolled Median Openings

Abstract

Uncontrolled median openings contribute to more road crashes because drivers do not obey traffic laws such as driving between lanes and following speed limits. This study examined the effectiveness of the different types of signs near an uncontrolled median opening in Mahmoud Abad, Iran. There were Signs 1 and 2 with a single message, while Signs 3 and 4 contained multi-messages. An Internet of Things (IoT) test system was developed to collect and record vehicle speed data before and after each sign installation. Results showed that speeding behavior decreased almost immediately after the road signs were installed and remained almost constant until the beginning of uncontrolled median openings. Moreover, multi-message traffic signs reduce vehicle speeds twice as much as single-message signs. Additionally, only multi-message signs are effective at reducing speed limit violations.

1. Introduction

At median openings, vehicles usually interact in a complex manner. Before entering an uncontrolled area, the driver must observe traffic flow, traffic signs, and applicable regulations. Upon approaching the U-turn section, turning vehicles (merging vehicles) should decelerate, stop when there is no gap to merge, and accelerate as soon as they start merging. As opposing traffic approaches the destination, it slows down to assist merging traffic, shifts to the right lane, or ignores the turning traffic. Traffic safety is more challenging when median openings are uncontrolled because U-turn movements are extremely dangerous and complex due to the high traffic volumes and the fact that approaching vehicles are traveling at higher speeds, making collisions more likely [1]. Further, because of impatient U-turn behavior that violates the priority rule, a median opening is also a risk-prone traffic facility. It is therefore necessary to prevent some risky driving behaviors in these areas, including speeding [2,3,4]. Due to the limited use of auxiliary lanes in these areas, some safety countermeasures may improve traffic safety to some extent. Traffic-calming devices are among the safety measures available [5,6,7]. The purpose of traffic calming is not to prevent the flow movement or to stop it, but in general, traffic calming is a set of measures and tools whose purpose is to reduce drivers’ risky behavior. In addition, traffic-calming devices are physical devices, such as barriers, lights, or signs, which physically slow down vehicles on streets or warn drivers to slow down on dangerous roads [5,6,7].

Previous research generally analyzed traffic-calming designs using three methods: a laboratory research environment (i.e., a simulator) [5,8], controlled environment (i.e., an instrumented vehicle) [9], and real-world environment (i.e., studying drivers’ natural behavior in the real world) [7]. Physical and perceptual traffic calming has also been studied extensively near intersection areas. A study examining the effect of speed cameras with and without speed warning signs showed that with the simultaneous use of cameras and signs, 4.5 km/h (2.8 mph) of speed was reduced and 2.5 km/h (1.55 mph) was reduced at the 85th percentile of speed limit without the speed camera [10]. Vehicle activated signs (VASs) and speed indicator devices (SIDs) have also been compared in a previous study [11], which concluded that SID signs reduce vehicle speeds more effectively than VASs on local routes. An earlier study investigated the effects of variable speed limit (VSL) signs on drivers’ behavior and perception, as well as text messages containing path risk content (such as low risk, medium risk, and high risk) [12]. The results showed that similar messages can have different effects depending on the situation. Another study examined the effectiveness of traffic-calming tools, speed tables, chicanes, and path slimming. Vehicle speeds were significantly reduced on average [9,13]. Research has been conducted on choosing safe gaps for U-turn movements near median openings [1,3,14], but none has examined the effect of traffic calming on speed changes or speed limit violations. Consequently, there is little information about the safety of uncontrolled median openings, a common intersection in many parts of the world, including the northern regions of Iran. To evaluate the effectiveness of IoT-based traffic calming and, at the same time, the safety of nearby areas, the present study aims to obtain a low-cost approach. The most important objective of this research is to determine which type of traffic-calming sign is most effective at reducing the frequency of speed limit violations as well as the speed selection behavior of drivers at different spatial distances.

2. Materials and Methods

2.1. Study Site

An uncontrolled median opening in Mahmoud Abad, Iran (Figure 1) was chosen as a case study since traffic police reports of this location and similar locations have shown an increasing trend in crashes in recent years. The location is a two-way separated road with three lanes on each side. There is a 30 m uncontrolled median opening in the road. During the summer of 2020, data were collected before and after traffic-calming devices were installed.

2.2. Advanced Speed Display

Passive infrared sensors (PIRs) have long been used in the field of transportation. For instance, these sensors were used in early studies [15,16,17] to control vehicular traffic. These tools have recently been used to develop new devices for real-time vehicle detection, classification, and speed estimation in wireless sensor networks [18,19,20,21].

In the present study, some environmental sensors were configured using the IoT to collect vehicle speed data. The advanced speed display is made by passive infrared sensors (PIRs), a minicomputer, and an interface that processes sensor data and displays the speed output as a variable message sign (VMS). Each pair of passive infrared (PIR) sensors measures the vehicle speed based on the time when a vehicle passes each sensor and the distance between sensors. The sensors are installed in pairs at different points along the route and are connected wirelessly using a modem to not interfere with moving vehicles and data collection. When a vehicle crosses a PIR sensor’s field of view, the intensity of the received infrared beam is reduced, which signals to the PIR the presence of a vehicle. The time at which the vehicle is detected is sent to the server (the minicomputer, Cubie board2), which calculates the speed using the simple relationship (x = vt) and time inputs from the pair of PIR sensors. Figure 2 shows the PIR speed recorders developed to collect data and a minicomputer (the main data processing server). The speed recording system is wirelessly connected to the display through the IoT. When a vehicle crosses the sensors, its speed is recorded by the system and displayed on the screen. The dimensions of the VMS used in the present study were 〖1.7〗 m × 〖2.5〗 m.

The road is monitored by PIR sensors positioned in pairs three meters from each other. Additionally, other paired sensors were placed 30 m from the prior paired sensors along the road. To detect the speed profile, which shows the driver’s response to an advanced speed display near an uncontrolled median opening, 6 points were considered (4 points before the median opening and 2 points after the median opening). At each of these points, the sensors were placed closely (with a distance of 3 m between them), at a distance of 30 m between them. In this distance, the measurement error had the narrowest margin, which was the optimal distance between pairs of sensors.

Each of the sensors at different case study locations/points was analyzed by several test vehicles as part of the validation process for the device used in the present study. Upon passing the sensor, the sensor recorded the speed at which the driver was traveling compared with the speed displayed on the speedometer of the vehicle. A camera mounted on the participants’ vehicle (CARPA-120 Dual Dashcam) filmed the inside and outside aspects of the vehicle. The playback resolution was 640 × 480 DVD quality, and the camera also captured interior audio. The vehicle-mounted camera also recorded GPS map data, including the exact location of the participant as well as the vehicle’s speed and acceleration rate. The speed recorded by each participant’s camera and the speed recorded by the sensors at 28 different points on the studied route (between point 1 and point 6) were analyzed. Based on the comparison of recorded speeds, it was determined that the sensor has an error of 1%.

The present study investigated the effects of traffic-calming methods using a before-and-after study approach. For this purpose, four scenarios were investigated (Figure 3), which are as follows:

• First scenario:

Sign 1: Installation of sensors on the road to display the vehicle speed on the VMS;

• Second scenario:

Sign 2: Installation of sensors on the road to display the vehicle speed on the VMS, and a speed limit sign (60 km/h);

• Third scenario:

Sign 3: Installation of sensors on the road to display the vehicle speed on the VMS, a speed limit sign (60 km/h), and a U-turn sign;

• Fourth scenario:

Sign 4: Installation of sensors on the road to display the vehicle speed on the VMS, a speed limit sign (60 km/h), a U-turn sign, and a sign guiding driving between lanes.

3. Results and Discussion

3.1. Speed Violation

Data collection took place on weekdays with dry pavement and good weather conditions. Data on the vehicle’s speed for each sign are in

Table 1. Speed limit violations and statistical tests of proportion on speed limit violations.

Sign	Total Volume (Number of Violations)		Violation Ratio (%)		Violation Average		Test Statistic (Zstat) and Significance at 95% Level b (SIG)
	Before	After	Before	After	Before	After	Zstat	SIG
1	(214) 5070	(199) 5219	4.22	3.81	42.8 a	39.8	0.14 b,c	✕
2	(248) 4826	(96) 5122	5.13	1.87	49.6	29.6	1.26	✕
3	(283) 4751	(42) 4895	5.95	0.85	56.6	10.4	1.98	✓
3	(259) 4505	(36) 4883	5.74	0.73	51.8	7.2	2.01	✓

^a Average value = 107/5 (days of date collection) = 21.4. ^b Z-critical at 95% confidence level (a = 0.05) is ±1.96. “✓“ indicates the significance and vice versa for “✕”. ^c Sample calculation to obtain the test statistic ($Z_{stat}$) is given as follows: proportion of speed violations (before), $p_1=\frac{214}{5070}=4.22\times {10}^{-2}$; the proportion of speed violations (after), $p_2=\frac{199}{5219}=3.81\times {10}^{-2}$; pool estimate of proportion, p = (214 + 199)/(5070 + 5219) = $4.01\times {10}^{-2}$. $\mathrm{Test}\mathrm{Statistics},Z_{stat}=\frac{p_1-p_2}{\sqrt{p\left(1-p\right)}\sqrt{\frac{1}{n_1}+\frac{1}{n_2}}}$ = $\frac{\left(4.22-3.81\right)\times {10}^{-2}}{\sqrt{4.01\times {10}^{-2}\left(1-4.01\times {10}^{-2}\right)}\sqrt{\frac{1}{5070}+\frac{1}{5219}}}=0.14$.

. From Saturday to Wednesday, data collection was conducted around the sites. Vehicle volumes vary within the days, but on average, 4000–6000 vehicles were recorded daily. Since Thursdays and Fridays are weekends in Iran, the speed of vehicles usually decreases on these days because of the high volume of traffic. As a result, no data were collected during these two days. Table 1 shows the traffic volume during the installation of each sign in before-and-after studies. It also demonstrates the number of speed limit violations in these periods. The evaluation of the number of speed limit violations in the two periods before and after the installation of the signs shows that the installation of the signs reduced these violations. When the first sign was installed, the average number of violations decreased from 42.8 to 39.8, representing a seven percent reduction. In other words, displaying the current vehicle speed on the VMS led to a change in the speed selection behavior of drivers. Therefore, the number of speed limit violations decreased by 7% after the installation of the first sign.

On the other hand, the average number of violations after the installation of the second sign was reduced by about 40%, which shows its greater effectiveness compared with that of the first sign. Table 1 shows that the average number of violations decreased significantly from 49.6 to 29.6. It is noteworthy that although both signs effectively reduce the frequency of speed limit violations, this effect is not significant at the significance level of 95%. Table 1 shows the test statistic values (Z stat) for both signs, which is less than 1.96. Table 1 also shows that the average number of speed limit violations decreased by 82% and 86% after the installation of Signs 3 and 4, respectively. The critical point is that installing these signs significantly reduced the number of violations at the 95% significance level. This shows that installing multi-message signs (Signs 3 and 4) has a far more significant effect than single-message signs (Signs 1 and 2) on speed limit violations. A section was chosen upstream to act as a control section, and the vehicles’ speeds were recorded. The goal was to consider any changes in the area that could affect the speed of the vehicles. There was no significant change in the speed of vehicles in the control section. Consequently, it can be inferred that the signs were responsible for the differences in vehicle speed at the study site.

3.2. Speed Profile

Each statistical period before and after the installation of the signs was analyzed in order to calculate the average speed, the 85th percentile speed, and the standard deviation of speed. SPSS software was used to calculate the required statistical metrics in this study. Statistical software SPSS was used to assess kurtosis and skewness for all data. A normal distribution is inferred from the obtained values of kurtosis and skewness, which are in the range of (2, −2). In

Table 2. Overview and comparison of vehicle speeds in various points.

Point	Study	Sign	Mean (km/h)	p-Value	H0
1	Before	-	65.89
	After	1	62.15	0.001	×
		2	63.84	0.005	×
		3	61.16	0.013	×
		4	59.68	0.004	×
2	Before	-	64.18
	After	1	62.91	0.012	×
		2	61.05	0.009	×
		3	59.18	0.016	×
		4	57.84	0.005	×
3	Before	-	66.75
	After	1	61.06	0.009	×
		2	60.85	0.005	×
		3	56.12	0.019	×
		4	55.51	0.014	×
4	Before	-	63.12
	After	1	60.29	0.000	×
		2	58.05	0.003	×
		3	51.55	0.002	×
		4	49.96	0.001	×
5	Before	-	64.18
	After	1	61.73	0.016	×
		2	58.30	0.011	×
		3	51.09	0.009	×
		4	50.00	0.010	×
6	Before	-	65.05
	After	1	63.13	0.022	×
		2	58.45	0.018	×
		3	51.13	0.009	×
		4	50.12	0.016	×

× indicates rejection of the null hypothesis.

, the average vehicle speeds are shown at different points. The results show a decrease in vehicle average speed. The deceleration for vehicles was observed at the beginning of the uncontrolled median opening after all signs were installed.

In order to evaluate the difference in mean speed before and after traffic calming, an independent t-test was performed. A t-test for independent samples compares the means of two independent groups in order to determine whether the associated population means are statistically different. This test assumes the average speeds are the same before and after traffic-calming measures (H0). An alternate hypothesis (H1) is that the mean speeds are different before and after the test. The null hypothesis can be rejected with a 95% level of confidence if the significance level of this test (determined by the p-value) is less than 0.05. Table 2 shows the results of t-tests using the data collected before and after the study. In Table 2, √ indicates confirmation and × indicates rejection of the null hypothesis.

Control locations were used to determine whether speed reductions were caused by the VMS or other exogenous factors. The control locations were chosen to reveal speeds at other locations without traffic-calming measures. At a distance of 100 m before the sign, the first control point is located, and at a distance of 100 m after the end of the median opening, the second control point is located.

Table 3. Overview and comparison of vehicle speeds at control points.

Control Points	Study	Sign	85th Percentile Speed	Mean (km/h)	p-Value	H0
1	Before	-	72.15	64.22
	After	1	74.24	66.75	0.415	√
		2	73.43	65.19	0.258	√
		3	72.85	67.09	0.335	√
		4	73.18	65.05	0.462	√
2	Before	-	74.85	66.72
	After	1	75.73	65.82	0.241	√
		2	74.54	66.35	0.119	√
		3	75.16	65.12	0.201	√
		4	76.55	65.20	0.505	√

√ indicates no rejection of the null hypothesis.

shows the average speed, 85th percentile speed, and speed changes for vehicles at the control points in the “before” and “after” periods. Based on the results of the t-test, the null hypothesis was not rejected, indicating that the average speeds were not statistically different. Meanwhile, the comparison of average and 85th percentile speed shows that drivers of vehicles crossed the spatial distances before the studied site at higher speeds in the “after” period. The results indicate that the signs on the road were responsible for the decrease in speed at the study points.

The speed profile of the vehicle passing from the site is shown in Figure 4. Before implementing the signs, drivers drove at speeds higher than 70 km/h (43.5 mph). Immediately following the installation of the signs, drivers started slowing down 30 m before the sign location. In general, the speed-reducing behavior continued until the beginning of the uncontrolled median opening, which is about 90 m. Afterward, the speed of vehicles remained constant for 30 m, representing the length of the median opening. After passing the endpoint of the median opening, the speed of vehicles increased again, although these changes are relatively small and gradual. Figure 4 shows that the installation of Signs 3 and 4 led to a greater change in the speed of vehicles compared to Signs 1 and 2. Changes in the speed of vehicles after the installation of Sign 4 are shown in the red line.

According to the collected data, the speed of vehicles gradually decreased from about 68.5 km/h to 61 km/h thirty meters before (the first point) the sign location, which indicates a sudden decrease in speed at this distance. Then, the speed reduction continued at a lower slope and decreased to 59.2 km/h and 55.6 km/h at points 3 and 4, respectively. The speed remained almost constant when passing the median opening with a length of 30 m, but it increased again after passing point 5 so that at point 6, it increased to more than 57 km/h. The speed at Signs 3 and 4 is almost the same, which can be seen in the orange line in Figure 4. The speed reduced from 69.5 km/h at point 1 to 58.12 km/h at point 4, which indicates a significant reduction. As with Sign 4, the reduction in the speed of vehicles due to the implementation of Sign 3 started from a distance of 30 m before the position of the sign and continued until the beginning of the opening (point 4). Speed changes after the installation of Signs 1 and 2 are shown in green and gray colors in Figure 4. Overall, Sign 1 made little change in reducing vehicle speed. The speed profile shows that the speed of vehicles reached from 69.5 km/h at point 1 to about 67.2 km/h at point 4 (the beginning of the median opening), which is not a significant reduction. On the other hand, a similar trend can be observed for Sign 2, where the speed of vehicles decreased from 70.7 km/h at point 1 to 64 km/h at point 4 (the beginning of the median opening), which shows a lower effect compared to Signs 3 and 4.

3.3. Discussion

Traffic safety at uncontrolled median openings is challenging since U-turning movements are complex and dangerous because of the high volume of traffic. In addition, approaching vehicles approach at a higher speed, increasing the risk of collision. Further, impatient U-turn drivers violate the rule of priority, making median openings a risky traffic facility. In the present study, signs that are relatively inexpensive were used, and the speed of vehicles near uncontrolled median openings was significantly reduced as a result. Road signs are commonly used due to their ease of use. The data in this study have clearly demonstrated the influence of signs. The effect of these signs was stronger close to median openings. Research has been conducted on modifying or combining some common signs to reduce vehicle speeds in different scenarios, which are mainly focused on areas near schools. In a simulation study, Akbari and Haghighi [5] found that modified signs decrease average traffic speeds. Their proposed modified signs include a warning sign for slowing down along with the speed limit. According to Saibel et al. [22], flashing speed limit signs effectively slowed cars down. Furthermore, Zhao et al. [23] found that flashing beacons, reduced speed signs, and crossing warning signs reduced speed. As demonstrated by Sun et al. [24], reducing the speed limit could improve traffic safety significantly.

The present study investigated the response behavior of drivers near the opening of the medians, which are mainly accident-prone locations, using multiple signs on an advanced display. Results revealed that the installation of Signs 3 and 4 significantly reduced the speed in the study area with 95% confidence. The trend of speed changes showed that the installation of Signs 4 and 3, which are a type of multi-message signs, reduced care speed by 13 and 11.5 km/h, respectively, before reaching uncontrolled median opening. On the other hand, car speed was reduced by 6.5 and 2.5 km/h at Signs 2 and 1, respectively, which shows a lesser effect than that of other signs. This indicates that signs and multi-message traffic signs can effectively reduce speed. Interestingly, the speed limit sign led to only a 6.5% speed reduction in the second scenario. Overall, the results of the four scenarios show that using speed signs alone cannot effectively reduce risky behaviors such as speeding. Still, to reduce these behaviors more effectively, it is necessary to use a set of traffic messages. Other studies have reported significantly better traffic safety when a combination of traffic-calming devices is used [6,7,9]. Although these studies have not investigated the changes in the speed of vehicles near an uncontrolled median opening, it seems that the combined traffic calming may lead to a further increase in traffic safety.

4. Conclusions

Safety on the road is one of the most pressing public health issues. Suitable solutions to this challenge can be found in modern and low-cost devices [25]. The use of IoT-based approaches to road safety is one of the new technologies. Additionally, traffic signs can reduce crashes without imposing additional costs through low-cost traffic-calming solutions. Our study investigated speed limit violations near uncontrolled median openings in Iran using an IoT-based low-cost approach. This study identifies and investigates speeding behavior near uncontrolled median openings in Mahmoud Abad, Iran. IoT-based calming devices were used to reduce vehicles’ speed in the four scenarios based on variable messages and traffic signs. With Sign 1, the speed of vehicles passing on the road was displayed. Sign 2 displayed the allowed road speed limit along with the permitted vehicle speed. The above two signs represent single-traffic message signs, while the third and fourth signs comprise multiple-traffic message signs. There is a U-turn sign along with the speed limit sign and the speed of vehicles passing through the road on Sign 3. Sign 4, in addition to the specifications of Sign 3, also includes a “Driving between Lanes” display sign. Results showed that the speed of the vehicles at multi-message signs, i.e., Signs 3 and 4, was about two times more than the speed limit display sign (Sign 2), leading to a speed reduction on roads. In spite of the fact that the speed limit sign (Sign 2) is effective in reducing speeding violations, it has a much smaller impact on vehicle drivers’ speeding behavior than multi-message traffic signs.

Future studies can investigate the limitations of the present study. One limitation of this study is that it does not examine scenarios at night. As a result, we cannot conclude what effect these signs will have regardless of the time of day. Moreover, while multi-message traffic signs were reported to be more effective at reducing speeding than single-message signs, the impact of these signs on driver perception and reaction time must be explored as well. Meanwhile, the distraction caused by spending time understanding the sign content may also represent a challenge in this regard. There is a possibility that this issue will become more important in areas where the speed of the vehicle is greater than the site in the present study since any delay in reaction time caused by distraction could result in more severe consequences if a crash occurs. Therefore, future work can include examining the scenarios in this research on high-speed roads or with different infrastructure conditions, such as more lanes. It is of course recommended to test this issue in simulation environments first, such as driving simulators or virtual reality devices, due to the importance of safety considerations [26,27].

To date, there is no global consensus regarding the safety level of autonomous vehicles (AVs). It is difficult to find common ground on a national or international level, aside from the global scale. A global consensus, however, has established that autonomous vehicles are not associated with driving fatigue, alcohol/drug impairment driving, or even lacking experience, which are all factors that may lead to or contribute to accidents [28]. NHTSA conducted a study for crash causation and examined whether accidents were related to recognition errors such as inadequate surveillance and internal/external distraction [29]. About one-third of accidents were related to decision errors such as driving misjudgment and false assumptions of others’ actions. In comparison with humans, it may be argued that AVs are less prone to error. Thus, they are capable of resolving such issues, and as a consequence, road accidents will be reduced. Due to these advantages, as long as AVs are comprehensively tested, and the test data are shared, processed, and used for improvisation globally, AVs will be more likely to be developed and their possibility to prevent accidents will rise.