Investigating the role of exiting vehicles and turn indicator usage in gap acceptance at single-lane roundabouts

Abstract

Roundabouts are becoming a preferred form of intersection control type due to their safety and capacity benefits. Yield control at roundabout entries inherently reduces fuel consumption and emissions when compared to other conventional intersection types. However, these benefits can be limited by poor driver behavior and judgment when entering the roundabout. This research addresses the nature of gap distributions and use of turn indicators by exiting vehicles at three single-lane roundabouts in Vermont, New York, and Alaska. Presented here is a comparison of vehicle headways measured at two different locations in each roundabout. Rejected headways are analyzed in the context of priority abstaining events when entering drivers yield to vehicles exiting on the same leg of the roundabout suggesting that “true” critical gaps are being overestimated. Results indicate that exiting vehicles, particularly those that do not use their turn indicators when departing from the major-stream of traffic have an influence on the entry decision of drivers on the same approach. This behavior is of particular concern for intersection efficiency (i.e., delay) and sustainability (i.e., excess fuel consumption and emissions). Results prompt the consideration of more consistent guidance on and enforcement of turn indicator use during roundabout negotiations.

Keywords

roundabouts / exiting vehicles / turn indicator use / headways

Introduction

Roundabouts are a particularly advantageous form of intersection type when considering the sustainability of transportation infrastructure. First and foremost, roundabouts serve to improve safety and capacity under most conditions. In addition, vehicular emissions and fuel consumption are also typically lower at roundabouts simply from yielding at entry rather than stopping as is required at other conventional intersection types. Other added benefits include reduced maintenance and related costs, aesthetics, and possible storm water retention in the central island. However, a roundabout must be properly designed and sized for these benefits to have the most impact. One particular challenge is accounting for and modeling the anticipated traffic (e.g., traffic volume and driver behavior characteristics) at an anticipated roundabout location. Without appropriately accounting for these characteristics, a roundabout design may not function as expected and the safety, capacity, and environmental benefits of the roundabout may be diminished. An important consideration in the design process is whether or not a driver decides to enter a roundabout based on other vehicles already circulating the roundabout.

The decision process of a driver at a roundabout entry requires the judgment of gaps between vehicles in the circulatory roadway. According to gap-acceptance theory, drivers will either: 1) enter the roundabout if there is a sufficient gap into which one can safely enter; or 2) stop if the gap is too small into which one could safely enter. In current methods, these gaps (both those that are accepted and rejected by entering drivers) are measured at the midpoint of the splitter island of the entry being observed. These measurements are used to determine critical gap which, along with follow-up time, are used to calibrate the 2010 Highway Capacity Manual [ 1] equation for roundabout capacity. Though the current headway method may be applicable for many drivers, it has been observed that some drivers do not adhere to the conventional yield-at-entry rule and stop despite there being sufficient gaps between circulating vehicles. It has also been noted that some entering vehicles will yield to vehicles exiting on the same leg that they may perceive to be vehicles that will continue circulating. In addition, vehicle trajectories through a roundabout are dynamic with respect to space and time (i.e., the gaps between vehicles do not remain constant as they traverse through the roundabout). In consideration of these issues, it is suggested here that the location where gap measurements occur should be farther upstream than where they are being measured currently.

This research addresses the issue of vehicle headway measurements in single-lane roundabouts. Using field data, this study focuses on a subset of driver behavior, referred to as priority abstaining [ 2], when entering drivers either: 1) stop when there are clearly sufficient gaps into which they could have entered; or 2) stop for vehicles that are exiting from the circulating stream of traffic on the same leg. This priority abstaining behavior directly reflects a decision process of drivers utilizing a reference point farther upstream than the splitter island. This behavior, as evidenced by vehicles that seem to “stop for no apparent reason,” may significantly influence the performance and safety of roundabouts. Video data of real-world traffic operations at single-lane roundabouts are used to investigate the distribution of vehicle headways in the context of priority abstaining. Since the distributions of headways are inherently changed by the presence of exiting vehicles, headways are analyzed at both the splitter island (conventionally used in most gap acceptance studies) and a secondary location where exiting vehicles can be included.

This paper seeks to address three objectives. First, circulating vehicle headways are compared at two locations within the roundabout to determine if there is a significant difference between them. Second, headways are compared when including exiting vehicles if a turn indicator was not used by the exiting vehicle. Finally, the differences in headways are then assessed in the context of priority abstaining events. The study locations included single-lane roundabouts in Middlebury, VT (Main St. & Cross St.), Glens Falls, NY (US 9 & Warren St.) and Fairbanks, AK (Helmericks Ave. & Herb Miller Blvd.).

Background

Because of the lack of strict priorty rules governing the interaction of major (i.e., circulating) and minor (i.e., entering) streams of traffic at roundabouts, driver behavior tends to be more varied. Commonly observed behavior types include more conservative-type behavior: priority sharing [ 3], priority surrendering, and priority abstaining [ 2]. Gårder [ 4] found that about one in 50 drivers who stop at the yield line did so when there were no vehicles to which they could yield. More aggressive-type behavior is evidenced by gap forcing [ 5], also referred to as priority taking [ 2]. Other research identifies priority modes as being when a roundabout is either operating under absolute priority mode, i.e., when minor stream vehicles have no affect on major stream vehicles, or under limited priority mode, i.e., when minor stream vehicles impede the major stream traffic flow [ 6].

Though the “rules at entry” for roundabouts are rather simple, the entry decision can be quite complex and augmented when entering drivers cannot discern between the maneuvers of exiting and circulating vehicles in the major stream of traffic. Mixed findings exist on the inclusion of exiting vehicles in capacity estimation of roundabout approaches [ 7, 8] with some research suggesting that the influence of exiting vehicles on entering vehicles is dependent on the distance between entry and exit points [ 9, 10]. This distance has been revisited in recent years yet there is still no agreed upon definition for the location of this point as it can be difficult to define where minor-stream drivers will not be able to identify major-stream drivers who are about to exit [ 1, 11]. This location can be dependent on variables such as the capability of the minor-stream driver, the speed and trajectory of the major-stream driver, and the size of the roundabout. Although exiting vehicle volumes have been explored as a potential variable in past research [ 12, 13], no conclusive findings exist that suggest how entry behavior relates directly to exiting vehicle volume. Here it is presumed that the use or turn indicators, rather the lack thereof, may help to explain the influence that exiting vehicles have on roundabout performance. A major-stream driver that does not use its turn indicator may influence the ability of a minor-stream driver to recognize the intent of the major-stream driver to exit and consequently affect the minor-stream drivers’ choice to enter.

Though some general recommendations can be found on correct navigation and use of turn indicators in roundabouts [ 14], more recent research shows variations on roundabout guidance, including the use of turn indicators, by state and no guidance specific to roundabouts within the 2000 Uniform Vehicle Code [ 15]. A recent case in the Court of Appeals of the State of Alaska [ 16] state that current regulations governing the use of turn signals (13 AAC 02.215) do not contain any provisions that expressly refer to roundabouts. Further discussion points to the fact that the existing regulation — in Alaska and the Uniform Vehicle Code [ 17] — requires that a motorist signals their intention to turn or move for not less than a continuous 100 feet before moving or turning to the left or right. This can be impractical or impossible to do in some roundabout configurations, though some publicly available information relating to specific roundabouts in Alaska indicate that “a right-turn signal is advisable when leaving the roundabout.” However, the ruling held that the regulations that prescribe rules for when motorists must signal their intentions were formulated before the state had roundabouts and therefore it becomes difficult, and potentially dangerous, to apply such rules at roundabouts. Further, variations in the blinking rate of turn indicators by vehicle make and model may also influence the ability of an entering driver to recognize the intent of some drivers to exit.

The consideration of exiting vehicles presents a significant issue when estimating the performance of a roundabout. As exiting vehicles leave the circulating stream of traffic they immediately alter the major stream headways (i.e., gaps). This is to say that headways being measured by an observer at the splitter island will not necessarily match the headways being judged by the entering driver who is using exiting vehicles in his/her decision process. This has implications for the estimation of critical gaps which are used in the calibration process of many regression-based roundabout capacity models, such as the capacity model for single-lane roundabouts in the 2010 Highway Capacity Manual [ 18]. Critical gap can be directly used to determine the capacity of a minor road junction or, in this case, the approach of a roundabout [ 19]. Brilon, Koenig, & Troutbeck [ 20] discuss more than 35 ways in which critical gap can be either defined or estimated and though there are a select few methods which are used with some frequency [ 21– 24], none on these deal with the issue of measuring gaps at location where vehicles exiting from a roundabout would be considered. It should be noted, however, that critical gap represents an average of the conditions under which it was measured and may vary based on traffic and geometric conditions [ 25– 28]. This research builds on previous work on non-compliant driver behavior at roundabouts [ 2] and focuses specifically on priority abstaining behavior, exiting vehicles, and the use of turn indicators of vehicles during their exit maneuvers.

Methodology

The objectives of this research are to determine if the location at which vehicle headways are measured produces different results, how much these measurements are affected when including exiting vehicles that do not use signal indicators, and if these differences matter in the context of priority abstaining events. For the analysis, it is hypothesized that drivers who exhibit priority abstaining behavior in the presence of an exiting vehicle are intrinsically using the “gaps” created between that vehicle and preceding/succeeding vehicles in their entry decision process. This is to say using the headway created by two circulating vehicles, as would be measured at the splitter island, would not be an accurate representation of the headways being judged by the entering driver. To test this hypothesis, data reduction methods were devised to obtain headway measurements and turn indicator use by exiting vehicles.

Data collection and reduction

Previously obtained video data obtained by Belz, et al. [ 2] and additionally collected data in Fairbanks, AK totaling 40 h of observations were reduced to obtain vehicle headways at two locations in each roundabout: 1) at the midpoint of the splitter island; and 2) at the halfway point between the analysis approach and the upstream approach (presumed to be an approximation of the diverge point between circulating and exiting streams of traffic). The Middlebury, VT, and Glens Falls, NY locations were selected from the previous study because they represented the two roundabouts with the observed highest and lowest proportion of priority abstaining behavior. All three roundabouts in this study had diameters of approximately 110 feet and though each roundabout had a different number of approaches. The approaches and areas selected for analysis at each roundabout (outlined in Fig. 1) controlled for as much between-site geometric variation as possible. As a result, the potential for variations in yielding rates to exiting vehicles due to differences in trajectories associated with exit radii and orientation is assumed to be negligible.

Vehicle headways and turn indicator usage by exiting vehicles were manually coded using a computer-aided script that timestamps a key-stroke command at a one-hundredth of a second resolution. For the headways at the splitter island, circulating vehicles were marked when the front of the vehicle crossed the midpoint of the splitter island. For headways at the diverge point, circulating and exiting vehicles were marked as the front of the vehicle crossed a superimposed line corresponding to the secondary point of reference. Exiting vehicles were also coded as having used their turn indicator or not. Observations that included heavy vehicles (e.g., tractor-trailer trucks, large RV’s, etc.), cyclists, and pedestrians have been removed from the analysis as each have their own unique effects on headways and the behavior of drivers around them. Consequently, this research presents only findings that relate to passenger vehicles.

Framework for including exiting vehicles

For the purpose of this study, the secondary location where headways were measured to capture exiting vehicles was chosen as the diverge point. The diverge point is defined as the location at which exiting vehicles begin altering their trajectory to exit on their intended leg of the roundabout. Since the trajectories of all vehicles are not identical, this location is simplified to the location where a line drawn perpendicular to both the central island and the external radius upstream of the approach in question intersects the midpoint of the circulatory lane (see point B in Fig. 2(a)). This is also assumed to be the point where a driver waiting to enter on an approach (represented as v₃ in Fig. 2(a)) is first able to discern between a vehicle that will continue circulating (v₁) and one that will exit on the same leg (v₂). Also for reference, point A in Fig. 2(a) represents the location at which vehicle headways are conventionally measured (at the midpoint of the splitter island) in roundabout analyses.

One of the shortcomings of gap acceptance theory is the general assumption of gaps in the major stream which remain unchanged [ 11]. In other words, the gap measurements obtained in the circulatory roadway will be identical regardless of the location at which they were measured. One case which violates this assumption is when minor-stream drivers force their way into small gaps after long wait times or under heavy circulating flows [ 6]. This behavior results in varied headway measurements between locations upstream and downstream of the merge point. Another violation occurs, as presented here, in the simple case when a major-stream vehicle exits the roundabout and immediately alters the original gap(s) between the leading and following circulating vehicles (see Fig. 2(b)). The new gap created will initially be the sum of the previously existing leading (h_o-h₁) and following gap (h₁-h₂). Once the circulating vehicles then reach the splitter island, it can be assumed that the trajectory of the leading vehicle has remained unchanged but the gap will be dependent on the nature of the driver who was following the exiting vehicle (v₃). More aggressive drivers may “close” the gap with the leading vehicle while more conservative drivers may “extend” the gap with the leading vehicle while they react to the exiting vehicle in front of them. In either event, the initial headway is replaced by one of two non-identical headways. For simplicity, the headway between leading and following circulating vehicles is generally assumed to remain unchanged [ 11]. In cases where multiple vehicles exit in succession (e.g., v_2,v₃ and v_n) the final cumulative headway would be from h_o to h_n+1.

Analysis

The two chosen roundabout locations were selected because of observed proportions of priority abstaining behavior. Figure 3 shows the percentages of observed turning movements and driver behavior. Percent exiting vehicles is expressed as those exiting on the leg being analyzed as compared to the total volume (circulating and exiting) at the diverge point. The Middlebury roundabout had a higher percentage of exiting vehicle traffic than did the Glens Falls roundabout. Similarly, the percent of exiting vehicles using turn indicators while exiting was higher at the Middlebury roundabout (53.5%) and Fairbanks roundabout (49.0%) than compared to the Glens Falls roundabout (36.9%). Percent priority abstaining behavior is referenced against the total entering vehicle volume. The rate of priority abstaining when the exiting vehicle did not use a turn indicator was similar at all three locations with 89.0%, 82.5%, and 81.6% occurring at the Middlebury, Glens Falls, and Fairbanks roundabouts, respectively.

Distribution of headways

For the headway analysis, only true gaps/headways are examined. Lags, the unexpired portion of a gap after the minor stream vehicle arrived at the entry point [ 22], were not considered in this study. In Fig. 4, you can see that the cumulative distribution of vehicle headways show much smaller headways (intuitively) when examining the mixed circulating and exiting vehicle streams. Note that the curves for the Middlebury, VT roundabout are not as smooth due to the lower sample size. Though all headways are included here, not just those accepted or rejected, only headways up to 12 s are shown as previous studies indicate that gaps longer than this are nearly always accepted and would be irrelevant for gap acceptance analysis [ 26].

This difference is illustrated further in Fig. 5 where vehicle headways at the splitter island are plotted against vehicle headways at the diverge point (which includes exiting vehicles). Clustering above the line of correlation for all roundabout locations provides an example of the new and larger gap created when exiting vehicles leave the circulating stream of traffic. Some of the remaining variation is presumed to result from the dynamic nature of vehicle trajectories while navigating through the roundabout (i.e., slower following vehicles creating larger gaps and faster following vehicles creating smaller gaps when comparing a downstream location to another point of reference upstream). This provides contradictory evidence to the supposition that the spatial distribution of circulating headways remains constant [ 11].

Distribution of rejected headways

Knowing that the headways are different by location, are influenced by exiting vehicles, and that entering drivers are influence by the exiting vehicles, it follows that the headways used in gap acceptance analysis should be adjusted or appropriately measured to account for these. Though having been considered previously using simulation [ 11], the effect and use of exiting vehicles in critical gap analysis is explored here using the empirically collected data. Since priority abstaining behavior is observable by the nature of rejected gap opportunities (ignoring cases where priority abstaining occurred when no vehicles were present), Fig. 6 characterizes the cumulative distribution of rejected headways at the diverge point and splitter island within the roundabout. Distributions are also shown for only headways rejected by vehicles exhibiting priority abstaining behavior. The upper limit of 8 s for rejected gaps corroborates previous work on the threshold of relevant gap values to be used in gap acceptance analysis at roundabouts [ 27]. The leftward shift in the distribution curves (though shown for rejected values<h would also be present in the distribution curve for values>h as would be used in Raff’s method for critical gap analysis) suggests that “true” critical gap of entering drivers is being overestimated when measured at the splitter island. In addition, one should note the similarity between the distribution curves for all rejected headways measured at the diverge point and rejected headways by priority abstaining vehicles at both roundabout locations.

Conclusions

Some priority abstaining behavior (i.e., excluding events when no vehicles are present in the roundabout at all) suggests that drivers are making judgments on gaps further upstream than where current practice has them being measured; this is evidenced by the yielding of entering drivers to exiting vehicles. Considering that nearly 88% of priority abstaining events at both roundabout locations occur during cases when exiting vehicles do not have their turn indicator activated may suggest that entering drivers are relying on other visual cues from drivers in the circulatory roadway other than solely relying on their own judgment of gap size. The role of turn indicator use on the gap decision process of entering drivers is presented here and it is evident that there is some between site variations in the use of turn indicators by drivers when exiting a roundabout (about one-half of drivers at the Middlebury and Fairbanks roundabouts and one-third of drivers at the Glens Falls roundabout). Although these findings are promising and provide insight into performance variations between roundabout locations, the way in which turn indicator use directly affects the entry capacity of roundabouts is not examined here and should be further explored. However, these results do provide some justification for policy changes and prompt consideration of more consistent state or federal guidance on the use and enforcement of turn indicators when exiting roundabouts.

The similarity between the distribution of rejected headways for priority abstaining vehicles and all rejected headways at the diverge point provides some evidence of where drivers are, in reality, making decisions on gaps in the circulatory roadway. These findings suggest that this point is indeed farther upstream than the splitter island and indicates merit in the use of the diverge point for future headway analyses. Further, the headways between circulating vehicles is shown to be dynamic when comparing measurements at the splitter island to those at the diverge point which counters the simplified assumption used in past research.

This research adds to the understanding of roundabout operations and entry driver behavior by using observations of real-world driver behavior and headways at roundabouts and represents an important consideration for the design and modeling of roundabouts that has yet to be comprehensively studied. The effect of exiting vehicles at roundabouts is a widely debated topic and though anecdotally accepted by many as a key factor in the gap acceptance process, rarely considered when modeling capacity of roundabouts. This research is not intended to provide a method by which to revise the use of critical gap in roundabout capacity analysis, yet provide additional evidence for the inclusion of exiting vehicles in current gap acceptance methodologies. Further, the use of turn indicators, or rather the lack thereof, has become an unfortunate driving culture norm. Without this critical information, it is difficult to distinguish the intentions of other drivers, inherently affecting the gap acceptance process and consequently the roundabout capacity.

The findings presented here are most applicable to smaller single-lane roundabouts where the distance between the point of entry and the decision point are relatively short. Results from this study also contribute to the existing database and knowledge related to drivers’ gap-acceptance behavior at roundabouts in the United States. Future work will seek to expand the roundabout sample size, include roundabouts from other regions in order to explore the use of turn indicators at roundabouts on a national level, and incorporate accepted headway distributions for a more comprehensive analysis of critical gap in the context of priority abstaining and exiting vehicles.

References

[1] Transportation Research Board. Highway Capacity Manual 2010. Washington, D C: National Academy of Science, 2010

[2] Belz N P, Aultman-Hall L, Garder PE, Lee BHY. An event-based framework for non-compliant driver behavior at single-lane roundabouts. Transportation Research Record No. 2402, Journal of the Transportation Research Board of the National Academies. 2014, C: 38–46

[3] Akcelik R. The relationship between capacity and driver behaviour. In: Proceedings of the 2nd International Conference on Roundabouts. Kansas City, MO: Transportation Research Board of the National Academies, 2008, 1–14

[4] Gårder P E. Little Falls, Gorham: Reconstruction to a Modern Roundabout. Transportation Research Record: Journal of the Transportation Research Board No. 1658, Transportation Research Board of the National Academies, Washington, D C, 1999, 17–24

[5] Kimber R. Traffic Capacity of Roundabouts, Laboratory Report 942. Crawthorne, Berkshire, England: Transport and Road Research Laboratory, 1980

[6] Troutbeck R, Kako S. Limited Priority Merge at Unsignalized Intersections. Transportation Research Part A, Policy and Practice, 1999, 33(3–4): 291–204

[7] Kimber, R., and Semmens, M.A Track Experiment on the Entry Capacities of Offside Priority Roundabouts. Crowthorne, England: Supplementary Report 334, Transport and Road Research Laboratory, 1977

[8] Brilon W, Stuwe B. Roundabouts in Germany: Recent Results Regarding Capacity and Safety. Nantes, France: In Actes du Seminar “Giratoir 92”, 1993

[9] Louah G. Recent French Studies on Capacity and Waiting Times at Rurual Unsignalized Intersections. Proceedings of Intersections Without Traffic Signals II, International Workshop. Bochum, Germany, 1988

[10] Simon M. Roundabouts in Switzerland: Recent Experiences, Capacity, Swiss Roundabout Guide. In: Proceedings of Intersections Without Traffic Signals II, International Workshop. Bochum, Germany, 1991

[11] Hagring O. Derivation of Capacity Equation for Roundabout Entry with Mixed Circulating and Exiting Flows. Transportation Research Record: Journal of the Transportation Research Board No. 1776, Transportation Research Board of the National Academies, Washington, D C, 2001, 91–99

[12] Mereszczak Y, Dixon M, Kyte M, Rodegerdts L, Blogg M. Including Exiting Vehicles in Capacity Estimation at Single-Lane US Roundabouts. Transportation Research Record: Journal of the Transportation Research Board No. 1988, Transportation Research Board of the National Academies, Washington, D C, 2006, 23–30

[13] Brilon W. Intersections without Traffic Signals, Vol. II. Berlin: Springer-Verlag, 1991

[14] Federal Highway Administration. Roundabouts: An Informational Guide. Mclean, VA: Turner-Fairbank Highway Research Center, 2000

[15] Rodegerdts, NCHRP Report 672: Roundabouts: An Informational Guide. Transportation Research Board of the National Academies, Washington, D C, 2010

[16] Curtis D. Noble v. State of Alaska. Court of Appeals of the State of Alaska, No. A-11041, Trial Court No. 4FA-10–4690 CR, No. 2473 – September 4, 2015

[17] National Committee on Uniform Traffic Laws and Ordinances. Uniform Vehicle Code. Alexandria, VA, 2000

[18] Wei T, Grenard J L. Calibration and Validation of Highway Capacity Manual 2010 Capacity Model for Single-Lane Roundabouts. Transportation Research Record: Journal of the Transportation Research Board, No. 2286, Transportation Research Board of the National Academies, Washington, D C, 2012, 105–110

[19] Ashworth R. The capacity of priority-type intersections with a non-uniform distribution of critical acceptance gaps. Transportation Research, 1969, 3(2): 273–278

[20] Brilon W, Koenig R, Troutbeck R J. Useful estimation procedures for critical gaps. Transportation Research Part A, Policy and Practice, 1999, 33(3-4): 161–186

[21] Drew D R. Traffic Flow Theory and Control. New York: McGraw-Hill, 1968

[22] Raff M. A Volume Warrant for Urban Stop Signs. Saugatuck, CT: The Eno Foundation for Highway Traffic Control, 1950

[23] Hagring O. Estimation of critical gaps in two major streams. Transportation Research Part B: Methodological, 2000, 34(4): 293–313

[24] Tian Z, Vandehey M, Robinson B W, Kittelson W, Kyte M, Troutbeck R, Brilon W, Wu N. Implementing the maximum likelihood methodology to measure a driver’s critical gap. Transportation Research Part A, Policy and Practice, 1999, 33(3-4): 187–197

[25] Tian A.A Further investigation on critical gap and follow-up time. In: Proceedings of the 4th International Symposium on Highway Capacity. Transportation Research Board of the National Academies, Washington, D C, 2000, 397–408

[26] Kittelson W K, Vandehey M A. Delay effects on driver gap acceptance characteristics at two-way stop-controlled intersections. Transportation Research Record, vol. 1320. Transportation Research Board, Washington, D C, 1991, 154–159

[27] Polus A, Shmueli S. Entry capacity at roundabouts and impact of waiting times. Road and Transport Research, 1999, 8: 43–54

[28] Xu F, Tian Z Z. Driver Behavior and Gap-Acceptance Characteristics at Roundabouts in California, Transportation Research Record. Journal of the Transportation Research Board, No. 2071, Transportation Research Board of the National Academies, Washington, D C, 2008, 117–124

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg