Sage Journals: Discover world-class research

Abstract

Objective

A driving simulator study was conducted to test the effect of directions pavement markings (DPMs) on drivers’ performance.

Background

The perception of traditional direction signs may be less efficient than that of DPMs because drivers allocate more attentional resources to the road than to any other location.

Method

Two experimental conditions of DPMs were tested: (a) the name of the lane’s destination town, inscribed in white letters; (b) red and white shields enclosing the name of the lane’s destination town. In both conditions, the name or shield was preceded by a white arrow and traditional road signs were presented as well. The experimental conditions were compared to a baseline condition, which consisted of traditional road signs only. Various scenarios of three-lane highways were simulated and the number of vehicles on the road was manipulated, forming a 2 × 3 design (two load conditions × three experimental conditions). Twenty-two participants took part in the experiment.

Results

Pavement markings in both configurations were highly beneficial. Compared with the baseline condition, in both experimental conditions, the drivers noticed the correct lane sooner and moved to the correct lane earlier. Results also suggested that white markings are more beneficial than shield markings; however, this finding should be further confirmed.

Conclusion

Lane DPMS are desirable as they are likely to improve traffic safety at relatively low cost.

Application

Regulations demanding the presence of DPMs, at the very least in low visibility decision sites, would increase traffic safety.

Keywords

attentional processes vision statistics and data analysis visual pictorial object displays driver behavior

INTRODUCTION

The current study claims that pavement markings designating important information, specifically regarding the lane direction, increase drivers’ perception, prevent unnecessary lane shifts, and improve traffic safety. Presenting important information via pavement markings consisting of written words or symbols may be very effective, on some occasions more than using traditional traffic signs solely. Since traditional signs are usually located on the side of the road (side signs) or above the road (overhead signs), they may be easily overlooked, as these locations are less attended while driving. Naturally, drivers allocate a vast amount of resources to the road, while only a small portion of resources is allocated to the sides or to the upper field of vision. Harbluk et al. (2007) found that drivers spent only three percent of the driving time looking at the sides of the road. Others found that many drivers failed to perceive specific road signs, and some signs were totally overlooked by all drivers (Ghasemi et al., 2020; Shoman et al., 2018). These findings confirm previous results that have been reported time and again, whereby drivers detect only a small portion of road signs (Costa et al., 2014; Milošević & Gajić, 1986; Shinar & Drory, 1983).

Kadar et al. (2011) found that most of the time drivers gaze at a distant point ahead, while aiming sporadic glances at the sides of the road only from time to time. When the vehicle speed increased from 30 to 110 kph, the point toward which the driver was normally looking became farther and the likelihood of the glances to the sides substantially decreased. Such findings suggest that all types of traditional road signs are harder to perceive at higher speeds. It can be assumed that side signs may be totally ignored at higher speeds. In addition, it can be assumed that the probability of missing overhead signs is increased as well because drivers’ attention is directed at a more distant point ahead, which may suggest that the ideal angle for perceiving these signs may occur when the sign is too far away to be read. This argument may explain the results of a study on overhead signs, in which participants made many unnecessary lane changes, changed to the correct lane at the last moment, or even missed the correct exit altogether (Upchurch et al., 2005). Likewise, the current study is focused on high speed driving in highway scenarios.

The most common usage of pavement markings, to designate the dividing line between lanes or road edge lines, have important known implications for road safety (Bektas et al., 2016; Donnell et al., 2009; Park et al., 2017). Arrow pavement markings defining the direction of each lane at decision sites are also prevalent, and shown to have a critical positive effect on traffic safety. Chrysler and Schrock (2005) reported a 90% reduction of wrong-way movements after such arrows were added at a freeway exit. Another study found a positive effect on road safety of arrow markings at roundabouts (Chong & Al-Mamari, 2020). Pour-Rouholamin et al. (2015) concluded that careful and accurate pavement markings in general, and particularly at problematic locations, can reduce the probability of collisions linked to inappropriate lane selection.

In many countries, pavement markings that provide information, aside from lane markings, edge markings, and arrows, are prevalent. The marked information may consist of safety or transportation instructions, for example, the words “Stop” designating a stop line or “Bus” designating a bus lane. Using pavement markings to designate information regarding route numbers or towns that can be reached via each lane (henceforth “direction pavement markings,” DPMs ) is also a common practice aimed at facilitating navigation. Colucci and Valdés (2017) found that DPMs, which started one km prior to an exit ramp, eliminated lane choice errors. Finley and Ullman (2011) compared three different DPMs conditions: name of the road to which the current lane leads within a colored shield, white arrows, and a combination of both name and arrows. The combined condition was found to be most effective. Ullman et al., (2010) concluded that the most beneficial design of DPMs is when the message is surrounded by shields, arrows precede these shields, and the markings are painted within the boundaries of the relevant lane.

The current driving simulator study explored whether adding DPMs would facilitate notice and choice of the correct lane (in other words, channel the driver to the correct lane). Three different conditions were tested: Baseline condition, which included only traditional direction road signs, White markings condition, which included traditional signs combined with repeated DPMs of the name of the town the lane leads to, written in white letters and preceded by a white arrow, and Shield markings condition, which included traditional signs combined with repeated DPMs of red and white shields containing the name of the town the lane leads to and preceded by a white arrow. The design of the shield was chosen based on its similarity to conventional shields designating the route numbers in Israel (e.g., Figures 1a and 2b). In addition, the study empirically tested the effect of the traffic volume (load) on the road on drivers’ performance. For this purpose, two different road load conditions were simulated: low and high (see details in Marciano & Yeshurun, 2015). Thus, using a driving simulator study enables performing a controlled exploration of the signs and pavement markings conditions along with controlled manipulation of load on the road.

Figure 1.

Different types of decision sites: (a) four-way intersection along a highway with three lanes in each direction; and (b) four-way intersection after a tunnel along a highway with two lanes in each direction. The yellow ellipses around the destination instructions and the yellow arrow near the correct lane in the signs were added here for Non-Hebrew readers.

Figure 2.

Experimental conditions examples. (a) Baseline condition: first sign at the right side, overhead sign as a second sign, high road load; (b) Baseline condition: two overhead signs, low road load; (c) White markings condition, low road load; and (d) Shield markings condition, high road load. The yellow ellipses around the destination instructions and the yellow arrow near the correct lane in the signs or pavement markings were added here for Non-Hebrew readers.

Both DPMs conditions were assumed to result in better safety measurements than the baseline condition. This hypothesis is based on the fact that the location of these DPMs (on the road) is more attentive then the locations of the traditional signs, and on their repetition along 1 km, before each turn. In addition, in line with former studies (Babić & Brijs, 2021; Ullman et al., 2010), it was assumed that the shield condition would be more effective than the white condition. Finally, a specific assumption was derived for intersections that appeared immediately after a tunnel. Due to the restricted visibility inside a tunnel, it was assumed that in these areas the benefit of both DPMs' conditions would be even greater.

METHOD

Participants

Twenty-two participants (12 women), all students at the University of Haifa, took part in the experiment in return for a monetary reward. The mean age of the participants was 28.2 years (ranging from 22 to 38, SD = 4.06). All the participants had normal or corrected-to-normal vision and driving experience of at least 5 years (mean of 9.9, SD = 3.6). All but two reported that they drive on regular basis. This research complied with the APA Code of Ethics and was approved by the Ethics Committee of the University of Haifa.

Tools

The study took place in a driving simulator using STISIM Drive® software on a PC computer. A Logitech steering system, which included a steering wheel and gas and brake pedals, was used. The participants sat on a stationary office chair, 2.5 m in front of a wide screen (2.3 × 3 m) subtending 62° of visual angle. A speaker (Studiophile BX5a Deluxe) providing background sounds was placed behind them.

Driving Scenarios and Task

General: Six different scenarios, each about seven km long, were developed. Names of non-existing towns of three letters each (written in Hebrew; using Arial font with fix size) were used as destinations. Using non-existing towns with short names was designed to eliminate the effects of familiarity and/or length of the name on the drivers’ perception. The scenarios simulated a highway with three lanes in each direction, separated by a road median (Figure 1), with a speed limit of 110 kph. Each scenario included few locations of four-way intersections with traffic light, in which the drivers had to decide and choose the correct lane. In some of the scenarios, a long tunnel of about one km, comprised of two of the three lanes, was also simulated near the scenario’s end. One-hundred meters after the tunnel exit, another four-way intersection with traffic light appeared (Figure 1b).

The name of the travel destination was presented in Hebrew throughout the entire scenario, in the lower left-hand corner (Figures 1 and 2), phrased as follows: “drive to ‘name of town’”. The participants’ main task was to drive to the specified destination while indicating, for each decision site in the scenario, the first time they observed a direction instruction designating the correct lane leading to their current target. To indicate the correct lane observed, they were asked to press one of six buttons located on the steering wheel (any one of three right buttons, to designate the right lane, and any one of the three left buttons, to designate the left lane). In addition, the participants were asked to move the vehicle to the correct lane (either by shifting lanes or by staying in the current lane if it was already the correct one) in order to prepare for the upcoming turn. Due to technical limitations of the software, the participants were asked not to actually perform the turns themselves. Instead they were ordered to continue straight ahead and look for the next direction instructions. Note that the measurements were detection time and distance of the direction instructions, as well as the time and distance of positioning the vehicle in the correct lane, and not the actual turning maneuvers. Therefore, the influence of this technical limitation on the current data is negligible.

Each of the six experimental scenarios was simulated in two different versions with regard to the vehicles load on the road. In the low road load condition, only three to four vehicles (on average) were presented at once in the vicinity of the driver in the driver’s headway lanes (i.e., in the driver sight ahead as well as in the mirrors; see Figures 1, 2b and c). In contrast, in the high road load condition this number was much larger, 10 to 15 vehicles (on average, Figure 2a and d). These vehicles were evenly divided between the three lanes. This experimental design resulted in 12 different combinations of scenario×road load.

Direction instructions: The development of the direction instructions (either traditional signs or pavement markings) was guided with the aid of an engineer with expertise in road engineering. Each of the 12 available scenarios had three different versions, according to the direction instructions condition: one control condition and two different experimental conditions, as will be described below. In the control (baseline) condition, the direction instructions included only traditional road signs, comprised of overhead and road side direction signs, designed as required by state regulations. Three different signs were presented prior to each intersection. A preliminary sign was placed 500 m before the intersection (first sign in Figure 2c), which did not indicate the direction of the lanes, but only designated an approaching intersection. The first direction sign was placed 250 m before the turn, and another was placed 70 m before the turn. In half the cases, the first sign was located at the right side of the road and the second was an overhead sign (Figure 2a) while in the other half both signs were overhead signs (Figure 2b).

In the experimental conditions , the traditional road signs were presented along with additional DPMs painted on the middle of each lane. These markings started 1,000 m before the turn, and were repeated every 70 m until the turn location. They included a direction arrow (right, left, or straight), followed by the name of a town to which the specific lane heads. Two different experimental conditions were tested: (1) white markings condition : all the markings were painted in white (Figure 2c), and (2) shield markings condition : the arrows were painted in white, but the names of the towns were written within a shield, similar to the shields containing route numbers customarily used within the traditional direction signs (Figure 2d).

The combination of six different scenarios, two road load conditions, and three types of conditions, yielded 36 different versions of the scenarios. This amount was necessary to prevent learning and familiarity effects, since each driver could drive to a specific destination in a specific condition only once. Finally, two similar practice scenarios with different town names, both including traditional road signs only, one with low road load and the other with high road load, were also developed.

Procedure

The participants took part in one session, which took about an hour and a half. After listening to the experimenter read aloud the instructions of the experiment, they drove in two practice scenarios. During the practice session, the experimenter verified that the participants pressed a button at the first time they encountered information regarding the directions and that they moved the vehicle to the correct lane, as instructed. Then, the participants drove in six scenarios of the baseline condition and were asked to take a ten-minute break before driving in twelve more scenarios of the experimental blocks. To control for fatigue or practice effects, the order of these two experimental blocks was counterbalanced between participants. Counterbalancing was also employed in the order of the scenarios within each block, ensuring that different participants would drive the scenarios in different order, and that low and high road load scenarios would be presented in alternate order.

DATA PROCESSING AND ANALYSES

Turns at intersections with traffic lights along the highway and turns at traffic lights that appeared after a tunnel were analyzed separately. The data for each combination of road load×experimental condition of each participant included twelve traffic lights along the highway and three traffic lights that appeared after a tunnel. The analyses averaged the responses of each participant to each of these two different decision sites.

Two measurements were analyzed for all cases: (1) Mean distance (m) between the location in which the driver pressed a correct button on the wheel and the turn itself (henceforth “ detection distance ”); (2) Mean time elapsed between the moment of pressing a correct button and arrival at the turn (henceforth “ detection time ”). Two measurements were recorded only if the driver shifted lanes; (3) Mean distance between the location in which the driver moved to the correct lane and the turn (henceforth “ in-lane distance ”); and (4) Mean time elapsed between the moment in which the driver moved to the correct lane and the turn (henceforth “ in-lane time ”). Note that in-lane distance and in-lane time are measurements that take into consideration not only the perception of the correct lane, but also the cognitive decision to move there and the actual moving reaction, which are also influenced by the traffic on the road (load). Note also, that since a maneuver to the correct lane was not always needed, since on some occasions the driver was already placed at the correct lane, the in-lane measurements involve missing data. In addition, measurements of time take into account the driver’s response as well as the vehicle’s speed, which is, of course, also controlled by the driver. For all measurements, higher values suggest that the drivers noticed the direction instructions or responded to them earlier, thus higher values reflect greater traffic safety.

Six different two-way general linear model analyses (GLMs), followed by least squared differences post hoc analyses (LSDs), were performed, one for each combination of decision location types and measurements, except for the two “in-lane” measurements of the “after tunnel” locations (due to too many missing data). All analyses included two within-subject variables: experimental condition (baseline, white, or shield markings) and road load (low versus high). Table 1 and Table 2 present the means, SDs, and range of each combination.

TABLE 1:

Descriptive Data: Means, SDs, Minimum and Maximum Values of the Different Experimental Conditions × Road Load Combinations of all Four Measurements at Intersections with Traffic Lights Along the Highway

Measurement	Condition	Mean	SD	Minimum	Maximum
Detection distance (m)	HL, baseline	343	78	125.9	668
	HL, shields	961	125	50.7	1025
	HL, white	1005	35	851.9	1047
	LL, baseline	345	65	52.0	627
	LL, shields	963	65	734.6	1021
	LL, white	1004	31	845.5	1045
Detection time (sec)	HL, baseline	20	13	5.4	71
	HL, shields	44	19	2.8	112
	HL, white	48	17	29.6	100
	LL, baseline	14	16	1.7	163
	LL, shields	34	5	25.6	49
	LL, white	36	5	27.8	53
In-lane distance (m)	HL, baseline	103	80	4.8	297
	HL, shields	544	249	18.3	882
	HL, white	632	244	64.3	932
	LL, baseline	117	69	9.2	273
	LL, shields	670	174	261.2	899
	LL, white	738	100	448.2	896
In-lane time (sec)	HL, baseline	5	5	0.4	21
	HL, shields	23	12	0.7	60
	HL, white	30	18	2.2	83
	LL, baseline	4	2	0.3	10
	LL, shields	23	7	8.5	37
	LL, white	26	6	0.4	40

Abbreviations: LL = low load; HL = high load.

TABLE 2:

Descriptive Data: Means, SDs, Minimum and Maximum Values of the Different Experimental Conditions × Road Load Combinations of Detection Distance and Time Measurements at Intersections After Tunnel

Measurement	Condition	Mean	SD	Minimum	Maximum
Detection distance (m)	HL, baseline	183	36	108.4	228
	HL, shields	883	255	200.7	1027
	HL, white	827	283	166.0	1012
	LL, baseline	169	33	91.5	215
	LL, shields	875	267	154.1	1022
	LL, white	841	287	121.2	1010
Detection time (sec)	HL, baseline	10	5	4.3	26
	HL, shields	47	23	10.2	106
	HL, white	48	26	12.7	95
	LL, baseline	6	1	3.2	8
	LL, shields	30	10	5.9	44
	LL, white	26	15	0.3	45

Abbreviations: LL = low load; HL = high load.

To overcome the missing data issue and further test the effect of the experimental condition on shifting to the correct lane, two-way GLMs tested the mean minimum values of in-lane measurements, for all sites together. Note that minimum value is a measure of the most extreme case, pertaining to the lowest distance or time of moving to the correct lane.

RESULTS

Intersections Along the Highway

Mean detection distance. The main effect of the experimental condition was significant [F (2, 42) = 3342.52, p < .0001; ηp² = 0.994]. All three conditions differed significantly from each other (p < .0001; Figure 3a). Compared with baseline condition, the distance was significantly increased by both DPMs conditions; however, the white markings further significantly increased the distance in comparison to the shield markings.

Figure 3.

Averaged mean of (a) detection distance (m); (b) detection time (sec); (c) in-lane distance (m); and d. in-lane time (sec) as a function of experimental condition at intersections along the highway. *—significant effect of the simple pairwise comparisons at p < 0.005.

Mean detection time. The main effect of the experimental condition was significant [F (2, 42) = 106.12, p < .0001; ηp² = 0.835]. The mean detection time in the baseline condition was significantly shorter than in the other two DPMs conditions (p < .008; Figure 3b), which did not significantly differ from each other. The main effect of road load was significant [F (1, 21) = 33.14, p < .0001; ηp² = 0.612]. The mean detection time was longer when the road load was high than low (37 sec vs. 28 sec, respectively).

Mean in-lane distance. The main effect of the experimental condition was significant [F (2, 42) = 316.94, p < .0001; ηp² = 0.938]. The mean in-lane distance of the baseline condition was significantly shorter than that of both DPMs conditions, with no significant difference between them (p < 0.005; Figure 3c). The main effect of road load was significant [F (1, 21) = 12.57, p < .0007; ηp² = 0.374], the mean in-lane distance was longer when the road load was low than high (508 m vs. 428 m, respectively).

Mean in-lane time. The main effect of the experimental condition was significant [F (2, 42) = 100.53, p = .0001; ηp² = 0.827]. The baseline condition was significantly shorter than the other DPMs conditions (p < .006; Figure 3d), with no significant difference between them.

Intersections After Tunnel

Mean detection distance. The main effect of the experimental condition was significant [F (2, 21) = 89.68, p < .0001; ηp² = 0.895]. All three conditions differed significantly from each other (p < .05; Figure 4a). Compared with baseline condition the distance was significantly increased by both DPMs conditions; however, the shields markings further significantly increased the distance in comparison to the white markings.

Figure 4.

Averaged mean of (a) detection distance (m); (b) detection time (sec) as a function of the experimental condition at intersections after a tunnel. *—significant effect of the simple pairwise comparisons at p < 0.05.

Mean detection time. The main effect of the experimental condition was significant [F (2, 42) = 45.35, p < .0001; ηp² = 0.683]. The mean detection time of the baseline condition was significantly shorter than the time for the other DPMs conditions (p < .05; Figure 4b), which did not significantly differ from each other. The main effect of road load was significant [F (1, 21) = 20.06, p < .0001; ηp² = 0.488]. The mean detection time was longer when the road load was high than low (35 seconds vs. 21 seconds, respectively). The interaction between the experimental condition and road load was also significant [F (2, 42) = 3.50, p < .05; ηp² = 0.143]. High road load resulted in larger detection times in all three conditions. However, the differences between the high and low load were larger for the white markings condition than for both the shield markings and baseline conditions (differences of 4 seconds, 22 seconds, and 16 seconds, for the baseline, white, and shields conditions, respectively; Figure 5).

Figure 5.

Averaged mean detection time (sec) as a function of the experimental condition and of road load at intersections after a tunnel.

All Intersections

In this analysis the measurement was the minimum in-lane distance or time (rather than the mean), a measure of the most extreme cases, that is, the lowest distance or time of moving to the correct lane, for each condition of each participant.

Minimum in-lane distance. The main effect of the experimental condition was significant [F (2, 21) = 77.73, p < .0001; ηp² = 0.787]. The mean minimum in-lane distance of the baseline condition was significantly shorter than that of the other conditions (p < 0.04; Figure 6a), which did not significantly differ from each other. The main effect of road load was significant [F (1, 21) = 15.42, p < .0001; ηp² = 0.423], the mean minimum in-lane distance was greater when the road load was low than high (411 m vs. 278 m, respectively). The interaction between the experimental condition and road load was nearly significant [F (1, 21) = 3.03, p = .0518; ηp² = 0.126]. A low road load resulted in larger minimum in-lane distances in all three conditions, but the differences between low and high road load was larger in the white markings condition compared to both shield markings and baseline conditions (differences of 14 m, 200 m, and 186 m for baseline, white, and shields conditions, respectively; Figure 7).

Figure 6.

Averaged (a) Minimum in-lane distance (m) and (b) Minimum in-lane time (sec) as a function of the experimental condition at all intersections. (*): significant effect of the simple pairwise comparisons at p < 0.05.

Figure 7.

Averaged mean in-lane distance (sec) as a function of the experimental condition and of road load at all intersections.

Minimum in-lane time. The main effect of the experimental condition was significant [F (2, 21) = 50.14, p < .0001; ηp² = 0.705]. The mean minimum in-lane time in the baseline condition was significantly shorter than in the other conditions (p < 0.04; Figure 6b), with no significant difference between them.

DISCUSSION

The effect of DPMs on noticing and choosing the correct lane was explored. Two different pavement marking configurations were tested: The name of the town that the lane leads to in a) white letters, or b) red and white shields containing the name of the town. In both conditions the name was preceded by a white arrow. These DPMs were always combined with traditional direction traffic signs, (roadside and overhead signs), and were compared with a baseline condition consisting of these traditional signs only. These conditions were tested with the aid of scenarios that simulated three lanes highway (in each direction), in which the road load was manipulated, yielding two conditions, high versus low (Marciano & Yeshurun, 2015).

In all analyses, both DPMs conditions showed greater values in comparison with the baseline condition. In their presence, drivers detected the correct lane sooner and moved to the correct lane earlier. The advantage of DPMs was further exhibited when only intersections after tunnel were investigated. In these cases, the mean detection time of the baseline condition was quite short and the mean minimum detection time was even shorter, suggesting that perhaps insufficient time is left for safe maneuver to the correct lane, especially when traffic on the road is heavy. This situation might lead to last minute lane shifts, which may increase the probability of accidents. In contrast, the mean detection time of the DPMs conditions was much longer (5 times longer), as was the mean minimum detection time (2 times longer). It can be concluded that the general beneficial effect of DPMs is even greater in the case of turns that appear immediately after a tunnel. Considering the low longitudinal and horizontal visibility within a tunnel, it is suggested that this DPMs advantage should be generalized to other low visibility conditions, such as night, rain, snow, or fog. Therefore, it is proposed that DPMs would be specifically beneficial in any low visibility condition, but this assumption should be empirically tested in the future.

In some cases, the measurements of white markings were significantly greater than those of the shield markings, and in other many cases this white advantage emerged but did not reach significance. Moreover, in some analyses the benefit of white over shields was further influenced by the road load. When only intersections after a tunnel were taken into consideration, shield markings improved performance compared with white markings, but when the interaction between road load and the experimental condition is considered, this pattern seems to be influenced by the load on the road. Drivers detected the lane markings sooner in the high load than in the low road load condition, but this simple effect was more pronounced in the white marking condition, influencing the main effect of the experimental condition.

The above discussion suggests that, contrary to the assumption concerning a benefit of shields over white markings, an opposite pattern, of advantage for the white over shields, was detected. This may be explained by the human visual system’s sensitivity to high contrasts (Wang et al., 2021), suggesting that white over black is more salient then red over black. This explanation fits the findings of a study reporting a benefit for white delineators over many other colors (Abou-Senna et al., 2021). In contrast, this explanation cannot be reconciled with former findings, in which, colored markings were more beneficial than white markings (Babić & Brijs, 2021; Ullman et al., 2010). It is suggested that the familiarity of drivers with a specific marking may explain the current finding. In the study conducted by Ullman et al. (2010) in the US, shield markings were most beneficial. While most drivers in the US are familiar with the common shield symbols that designate route numbers, this may not be true for the typical Israeli driver. It is important to distinguish between these two explanations, due to their different implications regarding the optimal pavement markings, therefore, this claim should be empirically tested.

LIMITATIONS

The limitations of the study offer opportunities for future studies. Due to STISIM driving simulator’s technical limitations, drivers could not perform real turns. Though this limitation had no practical effect on the current measurements, exploring the same question in a driving simulator that allows performing actual turns may add important insights. Additionally, simulator studies do offer close to real settings (Matowicki & Přibyl, 2017; Wynne et al., 2019), yet it is recommended to perform field studies to validate the results in a real driving setting.

CONCLUSION AND PRACTICAL RECOMMENDATION

According to the current results, it can be concluded that additional DPMs, consisting of either white or shields markings, would significantly improve traffic safety. They enhance earlier detection of the correct lane before arrival at a decision site and improve maneuvers to the correct lane, by channeling the driver to the correct lane, allowing sufficient time and distance to safely plan and execute the desired shift. This behavior leads to safer driving and prevent last minute lane shifts. Though the current findings also suggest white over shield advantage, this effect may be influenced by a familiarity effect of the shield, hence it may be inverted in other countries, where drivers are more familiar with route number shields (as was reported in former studies, e.g., Babić & Brijs, 2021; Ullman et al., 2010). This explanation should be further tested. To sum, DPMs are fairly cheap measures with evident added value for traffic safety. Promoting regulations demanding their presence, preferably in all decision sites but at the very least in low visibility decision sites, should be considered.

KEY POINTS

The effect of two different configurations of directions pavement markings (DPMs) on drivers’ performance was tested via driving simulator study.

Both configurations were highly beneficial, leading to earlier noticing and earlier movements to the correct lane in comparison with baseline (traditional signs) condition.

Lane DPMs are recommendable as they are likely to improve traffic safety at relatively low cost.

Footnotes

Acknowledgments

This work was supported by the Research Fund on Insurance Matters in Israel. The signs development was guided by eng. Yaakov Karni, road engineer expert.

Hadas Marciano Ph.D., Cognitive psychology, 2012, University of Haifa, Israel. Hadas Marciano is an organizational and cognitive psychologist, a senior researcher at the Institute of Information Processing and Decision Making, Ergonomic & Human Factors Unit, University of Haifa, Israel, and at the Stress and Resilience Research Center at Tel Hai Academic College, Israel. She also serves as a lecturer in the Psychology Department at Tel Hai academic College. Her research interests are human factors aspects of human behavior, and specifically of driver behavior, as well as human resilience (societal, community, and individual resilience).

References

Abou-Senna

El-Agroudy

Mouloua

Radwan

(2021). Effect of changing a traffic control device color on driver behavior and perception across different age groups. Transportation Research Record, 2675, 228–240. https://doi.org/10.1177/03611981211011168

Babić

Brijs

(2021). Low-cost road marking measures for increasing safety in horizontal curves: A driving simulator study. Accident Analysis & Prevention, 153, 106013. https://doi.org/10.1016/j.aap.2021.106013

Bektas

B. A.

Gkritza

Smadi

(2016). Pavement marking retroreflectivity and crash frequency: segmentation, line type, and imputation effects. Journal of Transportation Engineering, 142, 04016030. https://doi.org/10.1061/(asce)te.1943-5436.0000863

Chong

E. E. M.

Al-Mamari

R. N. S.

(2020). Characteristics of drivers’ lane choice at large multi-lane roundabout. International Journal of Integrated Engineering, 12, 176–183. https://doi.org/10.30880/ijie.2020.12.09.021

Chrysler

S. T.

Schrock

S. D.

(2005). Field evaluations and driver comprehension studies of horizontal signing (No. FHWA/TX-05/0-4471-2). Texas Transportation Institute, Texas A & M University System.

Colucci

Valdés

(2017). Phase III: Operational and safety-based analyses of varied toll lanes.

Costa

Simone

Vignali

Lantieri

Bucchi

Dondi

(2014). Looking behavior for vertical road signs. Transportation Research Part F: Traffic Psychology and Behaviour, 23, 147–155. https://doi.org/10.1016/j.trf.2014.01.003

Donnell

E. T.

Karwa

Sathyanarayanan

(2009). Analysis of effects of pavement marking retroreflectivity on traffic crash frequency on highways in North Carolina: application of artificial neural networks and generalized estimating equations. Transportation research record, 2103, 50–60. https://doi.org/10.3141/2103-07

Finley

Ullman

(2011). Field evaluation of in-lane pavement markings in advance of freeway interchanges. Transportation Research Record: Journal of the Transportation Research Board, 2250, 32–40. https://doi.org/10.3141/2250-05

10.

Ghasemi

Acerra

Vignali

Lantieri

Simone

Imine

(2020). Road Safety Review update by using innovative technologies to investigate driver behaviour. Transportation Research Procedia, 45, 368–375. https://doi.org/10.1016/j.trpro.2020.03.028

11.

Harbluk

J. L.

Noy

Y. I.

Trbovich

P. L.

Eizenman

(2007). An on-road assessment of cognitive distraction: Impacts on drivers’ visual behavior and braking performance. Accident Analysis & Prevention, 39, 372–379. https://doi.org/10.1016/j.aap.2006.08.013

12.

Kadar

E. E.

Rogers

S. D.

Costall

(2011). Gaze patterns in a steering-into-lane task on a straight road: The effect of driving speed, lane, and expertise. Ecological Psychology, 23, 93–122. https://doi.org/10.1080/10407413.2011.566034

13.

Marciano

Yeshurun

(2015). Perceptual load in different regions of the visual scene and its relevance for driving. Human Factors: The Journal of the Human Factors and Ergonomics Society, 57, 701–716. https://doi.org/10.1177/0018720814556309

14.

Matowicki

Přibyl

(2017). Cross-study research on utility and validity of driving simulator for driver behavior analysis. Acta Polytechnica CTU Proceedings, 12, 68–73. https://doi.org/10.14311/app.2017.12.0068

15.

Milošević

Gajić

(1986). Presentation factors and driver characteristics affecting road-sign registration. Ergonomics, 29, 807–815. https://doi.org/10.1080/00140138608968317

16.

Park

Abdel-Aty

Wang

J. H.

(2017). Time series trends of the safety effects of pavement resurfacing. Accident Analysis & Prevention, 101, 78–86. https://doi.org/10.1016/j.aap.2017.02.006

17.

Pour-Rouholamin

Zhou

Shaw

Tobias

(2015). Current practices of safety countermeasures for wrong-way driving crashes. In Transportation Research Board 94th Annual Meeting (No. 15-3648).

18.

Shoman

Simone

Vignali

(2018). Looking behavior to vertical road signs on rural roads. MOJ Civil Engineering, 4, 75–79. https://doi.org/10.15406/mojce.2018.04.00100

19.

Shinar

Drory

(1983). Sign registration in daytime and nighttime driving. Human Factors: The Journal of the Human Factors and Ergonomics Society, 25, 117–122. https://doi.org/10.1177/001872088302500111

20.

Ullman

B. R.

Finley

M. D.

Chrysler

S. T.

Trout

N. D.

Nelson

A. A.

Young

(2010). Guidelines for the use of pavement marking symbols at freeway interchanges: Final report (No. FHWA/TX-10/0-5890-1). FHWA/TX-10/05890-1. Texas A&M Transportation Institute, Texas A&M University System. http://tti.tamu.edu/documents/0-5890-1.pdf

21.

Upchurch

Fisher

Waraich

(2005). Guide signing of two-lane exits with an option lane: Evaluation of human factors. In Transportation Research Record 1918, TRB (pp 35–45). National Research Council.

22.

Wang

Tao

Liu

Zhou

(2021). Application of colour combinations on visual search tasks under vibration environments. The Journal of Navigation, 74, 311–327. https://doi.org/10.1017/s0373463320000533

23.

Wynne

R. A.

Beanland

Salmon

P. M.

(2019). Systematic review of driving simulator validation studies. Safety science, 117, 138–151. https://doi.org/10.1016/j.ssci.2019.04.004

The Effect of Lane Direction Pavement Markings on Driving Performance and Safety: A Driving Simulator Study

Abstract

Objective

Background

Method

Results

Conclusion

Application

Keywords

INTRODUCTION

METHOD

Participants

Tools

Driving Scenarios and Task

Procedure

DATA PROCESSING AND ANALYSES

RESULTS

Intersections Along the Highway

Intersections After Tunnel

All Intersections

DISCUSSION

LIMITATIONS

CONCLUSION AND PRACTICAL RECOMMENDATION

KEY POINTS

Footnotes

Acknowledgments

References