Abstract
The emergence of autonomous vehicles (AVs) presents significant transformative potential for modern transportation systems, particularly concerning pedestrian safety at roundabouts with heterogeneous traffic conditions and slip lanes. Slip lanes influence the dynamics of right-turning traffic by increasing vehicle speeds, reducing driver visibility of pedestrians, and altering merging behaviors, thereby elevating pedestrian conflict risks. (Note: This study applies to road systems where vehicles drive on the right-hand side.) This study investigates how different slip lane designs affect pedestrian safety, utilizing simulation modeling calibrated with field trajectory data. The findings indicate that vehicles in slip lanes typically maintain higher speeds than those in approach lanes, enhancing risk at entry and exit points. Notably, the presence of AVs significantly reduces pedestrian conflicts; with AV penetration at 75%, conflicts decrease by 34.62%, highlighting AVs’ potential to improve pedestrian safety. Moreover, the study provides targeted recommendations for slip lane configurations based on pedestrian activity levels: free-flow lanes are advised in areas with low pedestrian traffic to minimize delays while maintaining safety; yield-controlled lanes are preferred in medium-traffic areas for balancing flow and safety; and traditional roundabout designs are recommended for high pedestrian volumes because of their effectiveness in reducing pedestrian-vehicle conflicts. The study also emphasizes the importance of prioritizing safety enhancements, such as adjusting slip lane curvature and optimizing crosswalk positioning, particularly at entry sections of the slip lane where pedestrian exposure is highest.
According to the U.S. Highway Safety Manual, roundabouts have been shown to significantly reduce severe crashes resulting in loss of life by 78%–82% ( 1 ). This is attributed to their ability to enhance low operation speeds, which allows drivers more time to react to situations and reduces the severity of potential crashes. Additionally, roundabouts incorporate yield control at entry and have fewer conflict points compared with other types of intersection, further contributing to their safety benefits. However, it is important to note that not all road users, particularly pedestrians and cyclists who are considered vulnerable road users, may fully benefit from these proven safety advantages ( 2 ).
The emergence of autonomous vehicles (AVs) has ushered in a new era of transportation, offering potential benefits such as overcoming occlusions, reducing traffic congestion, improving fuel efficiency, and enhancing safety ( 3 – 5 ). However, the safety implications of AVs for non-motorized traffic need more investigation. Ensuring the safety of non-motorized traffic has become a critical concern, as it requires understanding the interaction between AVs and individuals who exhibit spatial and temporal non-compliance attitudes, perceptions, and behavioral norms that vary in different situations ( 6 , 7 ). Further exploration of this topic is crucial to assess how AVs can effectively interact with non-motorized traffic and accommodate their unique characteristics.
Moreover, besides the growth of AVs, the introduction of innovative roundabout designs such as flower roundabouts and turbo roundabouts, raises concerns that this safety advantage may not be universal ( 8 – 11 ). Infrastructural design parameters such as central island diameter, lane separations, crosswalk raising, and splitter islands have been found to affect pedestrian and cyclist safety ( 12 – 14 ). In addition to that, these newly developed roundabout designs have included exclusive right-turn lanes that lie adjacent to a roundabout to improve the roundabout performance and reduce conflicts ( 15 , 16 ). (Note: This study applies to road systems where vehicles drive on the right-hand side.) It is crucial to examine the implications of these design elements for the safety of pedestrians and cyclists within the context of the evolving transportation landscape.
Recent research has primarily focused on examining the safety of pedestrians and cyclists in roundabouts under mixed traffic conditions ( 17 , 18 ). However, there is little contribution made by these studies on the safety implications of AVs for pedestrians in roundabouts with slip lanes. The presence of slip lanes improves the level of service of the roundabout and reduces the conflicts at roundabouts because of diversion ( 15 , 19 ). However, slip lanes have a significant impact on the speed of right-turning traffic and vehicle merging behavior, and, as well, they raise the risk of conflicts for pedestrians. Therefore, further investigation is needed to assess the safety aspects of AVs for pedestrians in roundabouts with slip lanes, considering these unique challenges and their potential impact on pedestrian safety.
The primary objective of this research is to examine the safety implications of AVs for pedestrians (non-motorized traffic) in roundabouts. The study aims to provide a comprehensive analysis of how the presence of AVs affects pedestrian safety in roundabout slip lanes under mixed traffic conditions. This analysis involves assessing the rate of conflicts, the severity of conflicts, and the level of yielding compliance by both drivers and pedestrians as key indicators of pedestrian safety. The specific objectives of the study are as follows:
1) Investigate the impact of increasing AV presence in traffic on pedestrian safety at roundabout slip lanes using appropriate surrogate safety measures (SSMs) as indicators.
2) Examine the impact of slip lane designs on pedestrian safety at varying pedestrian activity levels.
By addressing these objectives, this study aims to enhance our understanding of the safety implications of AVs for pedestrians in roundabouts and provide valuable insights for transportation planning and design considerations.
The results of this analysis will increase knowledge of the impact of various slip lane designs on vehicle-pedestrian conflicts. Moreover, this study will identify an optimal slip lane design for improving pedestrian safety. Finally, the results will contribute to the existing body of knowledge on pedestrian safety in roundabouts and provide policymakers and transportation professionals with information on ensuring safe interactions between AVs and pedestrians in roundabouts.
Literature Review
According to the U.S. Pedestrian Safety Guide and Countermeasure Selection System (PEDSAFE), well-designed slip lanes should facilitate reduced vehicle speeds during turns, enhance visibility between drivers and pedestrians, and minimize pedestrian exposure to roadway risks ( 20 ). The literature outlines two design options for these slip lanes as seen in Figure 1. These include: 1) slip lanes with yield control at the exit approach and 2) slip lanes with an acceleration lane (free-flow lane). The warrant for slip lanes at roundabouts should be based on the level of right-turn flow rate, desired level of service at the roundabout, and non-motorized road user activity level in the area. Based on right-turn flow rate, Duan et al. recommended no slip lane for low flow, a yield-type slip lane for moderate flow, and a free-flow-type slip lane for high flow ( 21 ). Based on pedestrian activity level, the U.S. Federal Highway Administration (FHWA) generally warrants these slip lanes in locations with minimal pedestrian and bicycle activities, while advocating for yield-control in the slip lane on the basis of accommodating pedestrians and bicyclists because of the presence of a yield control ( 22 ). While the operational effects of slip lanes have been studied, their impact with regard to pedestrian safety at roundabouts remains underexplored ( 21 ).
Slip lane designs: (a) yield-control slip lane and (b) free-flow slip lane.
Most recent research on roundabout safety has investigated the different types of modern roundabout developed over the years. These include single-lane roundabouts, multi-lane roundabouts, turbo-roundabouts, mini-roundabouts, flower roundabouts, and roundabout interchanges ( 2 , 8 , 10 , 23 ). The specific type used may depend on factors such as traffic volume, available space, and design objectives. Despite the improvement of roundabout designs over the years that have improved their safety performance, the emergence of AVs has raised concerns about their safety as society is gradually transitioning between manual driving and autonomous driving ( 24 ). Although different studies have concluded that the presence of AVs reduces the number of conflicts and crashes at roundabouts, this benefit is dependent on the market penetration rate of AVs and the geometric design of the roundabout ( 5 ). Based on geometric design, modern roundabouts have been designed to incorporate right-turn bypass lanes that come with multiple operational benefits such as safety enhancement by reducing potential right-turning conflicts, as well as capacity improvement, and delay reduction for areas with heavy right-turn flow ( 25 , 26 ). According to Al-Ghandour et al., various types of slip lane exit, such as free-flow, yield, and stop, offer different levels of delay reduction, with free-flow slip lanes generally providing the most effective reduction in average delay ( 27 ). Based on the market penetration rate of AVs, most recent studies on non-motorized safety have made huge contributions to unveiling the impacts of AVs—there has previously been limited data on AVs. These studies have preferred the use of a traffic-conflict-based safety approach to analyze this impact on safety as it has been proven to have the potential to enhance safety analysis ( 28 ). Not only that, but also a simulation-based approach has been used by most studies, because of the lack of field data on AVs, and this has led to the development of assumptions when performing modeling ( 29 ).
Several studies have been conducted to determine the impact of different geometric features of a roundabout, employing either surrogate safety models such as the Surrogate Safety Assessment Model (SSAM) that utilizes surrogate measures such as traffic conflicts and near-misses to assess and predict safety outcomes, or using traditional approaches that utilize crash data ( 30 ). Notable parameters include, but are not limited to, the position of crosswalks, the presence of splitter islands, and entry/exit angles. Research suggests that factors such as crosswalk proximity and raised features influence pedestrian safety. The position of the crosswalk (flare) and raising a crosswalk have an influence on pedestrian safety at a roundabout. A crosswalk near the roundabout will increase the probability of rear-end crashes and queue spillback into the roundabout. Meanwhile, a crosswalk further away from the roundabout will encourage exiting drivers to accelerate, therefore influencing pedestrian safety ( 31 ). In a quasi-experimental before-and-after study conducted by Candappa et al., the effectiveness of raised crosswalks at roundabout entrances in improving pedestrian safety was examined ( 13 ). The results indicated a significant decrease in the average speed of approaching vehicles, and a perception of enhanced convenience and safety among pedestrians compared with the pre-treatment period. Another significant roundabout design parameter that influences pedestrian safety is the splitter island. According to research, splitter islands tend to lower the potential for pedestrian-vehicle conflict ( 14 , 32 ). This is attributed to the extra pedestrian holding space in the splitter island that has a significant role in determining pedestrian crosswalk performance ( 33 ). To account for this extra pedestrian holding space, studies have focused on analyzing the impact of the width of the splitter island and its correlation to roundabout performance parameters such as stop-line delay and critical gap ( 34 , 35 ). In addition to that, entry and exit angle parameters have an impact on pedestrian safety at roundabouts. A recent study by Novák et al., determined that entry design parameters have a statistically significant influence on safety as far as crash frequency, severity, and speeds are concerned ( 36 ). According to Weber, a more tangential exit will allow higher speeds through the exit crosswalk ( 31 ). Also, a tangential exit angle will provide pedestrians with an early determination of whether a circulating vehicle will exit or continue circulating.
Despite the considerable research effort to analyze pedestrian safety by evaluating various roundabout design parameters, there remains a notable gap in the literature about the impact of slip lanes on pedestrian safety at roundabouts. Slip lanes are perceived to improve the level of service as well as reduce conflicts ( 15 , 19 ). However, it is important to note that the safety benefits associated with slip lanes may vary depending on the specific design characteristics and the presence of AVs. Designs typically prioritize motorists, potentially reducing the visibility of pedestrians. Drivers often concentrate on navigating roundabout traffic, potentially overlooking pedestrians. Therefore, there is a need to investigate the safety implications of different slip lane designs and their interaction with AVs to better understand their effects on pedestrian safety at roundabouts.
Methodology
Research Flow
This study aimed to determine the impact of slip lanes on pedestrian safety using a simulation approach and SSAM. The process consists of four main processes: 1) field data processing, 2) simulation with VISSIM under different AV scenarios, 3) calibration of the VISSIM model, and 4) comparative analysis of the slip lane designs using SSAM conflict analysis. Figure 2 shows a conceptual framework of the steps followed to carry out this study.
Impact analysis framework: step-by-step process.
Site Location and Field Data
For data collection purposes, a roundabout with a slip lane would be an ideal location for analysis. Therefore, the intersection of State St. and W. Ellsworth Rd. in Ann Arbor, Michigan, U.S., shown in Figure 3, was selected as a representative for analysis. Field trajectory data recorded at this roundabout for September 1–12, 2022, was retrieved from a recent study done by Zhang et al. (available on GitHub) ( 37 ). The roundabout has four approaches with the southbound (SB) approach having a yield-control slip lane for right-turning traffic. In this study, trajectory data for the afternoon peak period from 2:00 to 5:00 p.m. were retrieved and used for calibration of the VISSIM simulation model.
Google Earth photo of the roundabout at the intersection of State St. and W. Ellsworth Rd. in Ann Arbor, Michigan, showing trajectories analyzed in this study.
Trajectory Speed Data Analysis
To analyze the drivers’ speed behavior while approaching the entry and exit pedestrian crossings, as illustrated in Figure 1, speed data from the trajectory data was obtained for three trajectory paths (i.e., northbound entry, westbound entry traffic, slip lane traffic) as shown in Figure 4. The study retrieved speed data from the 2:00 to 3:00 p.m. peak period. The purpose of this analysis was to assess how the drivers’ speed adaptation while approaching a pedestrian crossing was affected when driving on a slip lane compared with other approaches.
Vehicle trajectories evaluated in this study.
Simulation and Conflict Analysis
To evaluate the safety impact of AVs and simulate various slip lane designs, this study utilized VISSIM as the traffic simulation platform. Additionally, SSAM was employed to extract the number of potential conflicts using SSMs from the simulated trajectory data. The subsequent sections provide details on the development and calibration of the VISSIM model, as well as the extraction of SSMs.
VISSIM Modeling and Calibration
This study employed VISSIM version 23 to create simulation models for vehicles and pedestrians for the roundabout at the intersection of State St. and W. Ellsworth Rd when having 1) yield-control slip lane and 2) free-flow slip lane as shown in Figure 5. Field data obtained (as described in the Site Location and Field Data section) was used for the model calibration. In our calibration process, specifically, we selected 55 complete trajectories for each of the 16 movements within the roundabout. This selection was based on the originating and destination lanes of each movement, resulting in a total of 110 trajectories per entry point (two lanes per entry point, except the slip lane). The free-flow slip lane model was developed after calibrating the yield-control slip lane model. Although the specific operational characteristics between yield-control and free-flow slip lanes (such as the geometric characteristics and the yield-control) may differ, certain fundamental elements, such as vehicle speeds, trajectories, and conflict points, may share similarities. Leveraging the data collected from the yield-control slip lane allows for the development of a calibrated model that approximates the behavior of the free-flow slip lane within the simulation by altering the geometric properties of the slip lane and removing a yield-control.
VISSIM simulation model for State St. and W. Ellsworth Rd roundabout: (a) with yield-control slip lane and (b) with free-flow slip lane.
The choice of car-following model utilized for the human-driven vehicles (HDVs) and AVs were the Wiedemann 74 model and Wiedemann 99 model. The base model was calibrated using the Wiedemann 74 car-following model. In comparison with HDVs, AVs have a more assertive behavior with shorter headways and aggressive acceleration and deceleration behavior, and implement multiple vehicle cooperation and collision avoidance because of connected vehicle technology ( 18 ). These behaviors of the AVs are reflected by modification of the behavior models, as shown in Table 1, with shorter standstill distance, shorter safety distance (lower headway), and following variation. Negative following threshold and positive following threshold are set to smaller values to mimic the sensitivity of AVs when following. Speed dependency of oscillation is set to zero to portray AVs’ ability to follow without any oscillation. Also, a more aggressive acceleration and deceleration are expected for AVs. Therefore, these parameters have been adopted in this study, as shown. The default values are from VISSIM, while the HDV and AV parameters are adopted from the study by Granados et al., ( 38 ). These parameters were adopted by different other recent studies as well ( 5 , 39 , 40 ).
Parameters for Human-Driven Vehicle (HDV) and Autonomous Vehicle (AV) Behaviors
Note: na = not applicable.
Surrogate Safety Assessment Model (SSAM)
SSAM was used in this study to analyze SSMs using simulation data. Trajectory files generated by the VISSIM traffic microsimulation program were analyzed in SSAM to identify potential conflicts occurring at the entry and exit crossing point in the slip lane, as shown in Figure 1 ( 18 , 30 ). Although SSAM does not directly distinguish between pedestrian-vehicle conflicts, it provides the identification of vehicles involved in conflicts.
To identify pedestrian-based conflicts, two criteria were applied. Firstly, conflicts with an angle greater than 80° (default value for crossing-type conflict in SSAM) were considered crossing-type conflicts. Various studies propose different angle thresholds (such as >85° and >60°) for vehicle-vehicle interaction ( 41 , 42 ). However, for vehicle-pedestrian or bicycle interaction, studies have adopted a similar angle used in this study (i.e., >80°) ( 43 , 44 ). This angle range accommodates variations in the interaction angle between pedestrian and vehicle directions, as not all pedestrians will approach the crossing point from a direction perpendicular to the flow of the vehicular traffic, and also this criterion takes into account instances where drivers may switch lanes within the vicinity of the crossing point. Secondly, the shorter length of pedestrians (0.3–0.5 m) compared with vehicles (>3 m) allowed for the distinction of pedestrian-based conflicts. By extracting vehicle IDs and types from VISSIM, the number of potential conflicts categorized by vehicle types (HDV-pedestrian and AV-pedestrian) was determined.
SSAM utilizes two threshold values, time-to-collision (TTC) and post-encroachment time (PET), to identify potential conflicts. While default values are commonly used for vehicle-vehicle conflicts, adjusted threshold values were applied in this study to enhance the detection of pedestrian-vehicle conflicts. A TTC threshold of 2.7 s and a PET threshold of 8 s, based on previous research, were employed to capture conflicts involving pedestrians ( 45 ).
Comparative Analysis
The study aimed to make a comparative analysis to evaluate the impact of slip lane design on pedestrian safety while considering the effect of AVs. To make this comparison and evaluate the impact of AVs, varying penetration rates of AVs were analyzed for each of the two cases of slip lane design ( 5 , 18 ). The AV penetration rates analyzed were 0%, 25%, 50%, and 75%, where the scenarios were defined as follows:
0%: 98% HDVs, 2% heavy goods vehicles (HGVs), 0% AVs (base)
25%: 73% HDVs, 2% HGVs, 25% AVs
50%: 48% HDVs, 2% HGVs, 50% AVs
75%: 23% HDVs, 2% HGVs, 75% AVs
The impact of slip lanes on pedestrian safety was compared between HDVs and AVs using SSMs. the study utilized SSMs—TTC and PET—leveraging SSAM which provided filtered conflict files for detailed vehicle-pedestrian interaction analysis. This approach enabled the exploration of how AVs and HDVs differently influence safety metrics within each slip lane design. Based on the work by Miqdady et al., the second vehicle involved in a conflict is considered the decision-maker vehicle that potentially can either avoid or contribute to the conflict ( 24 ). Therefore, an equation was adopted to calculate the involvement ratio of the second vehicle in potential vehicle-pedestrian conflicts. This involvement ratio equation is represented by Equation 1:
where
Vt = the vehicle interaction (e.g., HDV-pedestrian with the HDV as the second vehicle) in a conflict, and
i = the scenario ranging from 0% to 75%.
The impact of each slip lane design on pedestrian safety was assessed by comparing the speed profiles of traffic moving through the slip lanes with those at a normal roundabout approach. This comparison was further extended to analyze the mean values of SSMs across the different AV scenarios, factoring in varying pedestrian activity levels, to provide a nuanced understanding of how slip lane design influences safety at roundabouts and develop recommendations.
Results and Discussion
The findings of this study are presented in two distinct sections. Firstly, the impact of AVs on pedestrian safety is examined by conducting a comparative analysis of SSM values between HDVs and AVs. A t-test is employed to assess the significance of differences in SSM values across four scenarios generated in VISSIM. The results of this section provide valuable insights into the specific SSM indicators that are significantly affected by the presence of AVs, aiding in the interpretation of their impact on pedestrian safety. Secondly, a comprehensive comparison of slip lane designs is conducted using the same four scenarios. The analysis involves two components. Firstly, the SSM indicators of the different slip lane designs are compared to identify any variations in safety performance. This examination allows for a thorough evaluation of the safety implications associated with each slip lane design. Secondly, the study examines how different slip lane designs influence pedestrian safety at roundabouts across various levels of pedestrian activity. This comparison provides a quantitative assessment of the safety outcomes associated with each slip lane design.
Impact of AVs on Pedestrian Safety at Slip Lanes
Before performing the t-test, an assessment was conducted to examine the distribution of the SSM data and assess the homogeneity of variances. To evaluate the normality of the data, both Shapiro-Wilk tests and Kolmogorov-Smirnov tests were employed. The results from these tests indicated that the SSM data followed a normal distribution. Thus, meeting the assumption of normality, which is essential for conducting a reliable t-test.
SSM Analysis across AVs Scenarios and Vehicle Category
The one-tailed independent sample t-test was used to compare the TTC and PET mean values between AV- and HDV-pedestrian conflicts in each of the four scenarios generated while assuming unequal variances. Results revealed that HDVs exhibited a superior TTC mean value compared with AVs in two scenarios, namely 50% and 75%, as presented in Table 2 and Figure 6. Moreover, as the presence of AVs in the traffic increased, the TTC value for HDVs increased, while the TTC value for AVs decreased. This could be attributed to the influence of slip lanes, which encourage higher speeds and subsequently iaffect the TTC values for AVs, considering their inclination toward higher speeds compared with HDVs, as described in this study. However, a statistically significant difference in TTC mean values between AVs and HDVs was only observed in the 75% scenario (p = 0.0336). With regard to PET, both AVs and HDVs exhibited a decrease in mean values from the 25% to the 50% scenarios. Likewise, HDVs showed a decrease from the 0% to the 25% scenarios, followed by an increase in the 75% scenario. A low TTC value typically indicates that there is little time for a collision to be avoided between a vehicle and a pedestrian at a slip lane crosswalk. As the AV penetration rate increases, a declining trend in PET and TTC at roundabout slip lanes underscores the need for optimal crosswalk positioning and visibility enhancements between pedestrians and vehicles. This can be achieved by adjusting slip lane curvature or alignment to ensure clear lines of sight. Existing guidelines recommend a compound curve radius with a larger radius preceding to slow vehicle speed and enhance driver vision ( 46 ). Additionally, placing crosswalks further from slip lane entries reduces conflict risk, allowing vehicles more space to adjust speed after detecting pedestrians. Crosswalk placement guidelines suggest centering it at the slip lane for both yield-controlled and uncontrolled lanes; however, placing it downstream for yield control enhances pedestrian safety ( 45 ).
Significance Test on the Difference of Autonomous Vehicles (AVs) and Human-Driven Vehicles (HDVs) Surrogate Safety Measures (SSMs) Mean Values
Note: PET = post-encroachment time; TTC = time-to-collision; na = not applicable.
p-value < 0.05 is considered statistically significant.
Surrogate safety measure (SSM) mean values for pedestrian-vehicle conflict at a yield slip lane roundabout: (a) time-to-collision (TTC) mean values and (b) post-encroachment time (PET) mean values for each scenario.
Conflict Analysis across AVs Scenarios and Vehicle Category
The results presented in Table 3 reveal a consistent pattern: as the penetration rates of AVs increase across from the 0% to the 75% scenarios, there is a notable decrease in the total number of pedestrian conflicts. This finding aligns with previous studies that have examined the relationship between AV penetration rates and conflict occurrence ( 5 , 18 , 24 ). However, the magnitude of the reduction becomes more pronounced with higher AV penetration rates, particularly in the fourth scenario, where a substantial decrease of 34.62% is observed. In addition to that, the analysis conducted using an analysis of variance test demonstrates that the scenarios are statistically significantly different from each other at a 95% confidence level. This statistical evidence further supports the notion that the variations in AV penetration rates across scenarios have a significant impact on the total number of pedestrians. Furthermore, Equation 1 was used to evaluate the involvement ratio of the AV and HDV at-fault in every conflict evaluated. As illustrated in Table 3, there was an observed decrease in the involvement of AVs as the at-fault party with the increase in penetration rate, while there was a corresponding increase in the involvement of HDVs. Recent research has suggested information-sharing methods and local collaborative information learning aimed at enhancing AV awareness and responsiveness to challenging scenarios, such as those encountered in slip lanes ( 3 , 4 , 47 ). By leveraging information sharing capabilities among AVs, the system can significantly reduce pedestrian conflicts at slip lanes by enhancing detection, awareness, coordination, and responsiveness to pedestrian movements ( 4 ).
Number of Conflicts by Scenario and Conflict Type on a Yield-Control Slip Lane
Note: AV = autonomous vehicle; Avg = average; HDV = human-driven vehicle; Involv. = involvement ratio; Ped = pedestrian; values in brackets = standard deviation; na = not applicable.
Impact of Slip Lane Designs on Pedestrian Safety
This section delves into the effects of slip lane designs on pedestrian safety at roundabouts, with a focus on three key aspects: speed profiles, SSMs, and the influence of varying pedestrian activity levels.
Speed Profiles at Roundabout Slip Lanes
Figure 7 illustrates an interquartile range plot of speed distribution every 10 ft for the slip lane traffic (Figure 7a) and SB traffic (Figure 7b) with distance measured from the entry crossing point. The x-axis indicates the 10 ft markers along the direction of travel toward the crossing point, while the y-axis shows the speed (in mph). The horizontal line shows the mean speed value at that distance marker. It is worth noting that the speed distribution of traffic on the slip lane (mean speed at 0 marker = 6.8 mph) is considerably higher than for the SB traffic (mean speed at 0 marker = 1.8 mph) at the entry location. This is similar to the article written by Davis, condemning the existence of slip lanes because of prioritization of speed over safety ( 48 ). This disparity can be attributed to SB traffic decelerating before merging into the roundabout and the placement of a crosswalk near the roundabout entry to ensure safety. Meanwhile, slip lane traffic perceives the slip lane as an exit from the roundabout and chooses to accelerate toward it. However, this behavior has a great impact on pedestrian safety at roundabouts featuring slip lanes.
Interquartile range (IQR) plots showing speed profiles for traffic flow at the entry location as shown in the diagram (left): (a) IQR plot of the speed distribution for the slip lane traffic, (b) IQR plot of the speed distribution for the southbound (SB) traffic.
Furthermore, Figure 8 provides an opportunity to understand the speed patterns of the slip lane traffic after crossing the entry crossing point and before crossing the exit crossing point. A decrease in speed can be observed at the region from the 150 ft marker to 90 ft marker. This is because of the presence of yield control at the merge area. Moreover, the speed distribution of traffic from the yield-control slip lane is much lower compared with traffic from the roundabout approaches at the exit. This can be attributed to certain roundabout parameters, such as the design of the exit (including a tangential exit) which encourages drivers from the roundabout to accelerate toward the exit. These speed patterns at the exit influence the pedestrian yielding behaviors, as pedestrians are more inclined to yield to traffic at the exit of a roundabout compared with the entry of a roundabout ( 49 ). However, pedestrian yielding behavior at slip lane entry locations can be influenced by the type of slip lane. Consequently, safety improvements, such as slip lane curvature combination and crosswalk placements, should primarily target the entry section of slip lanes where pedestrian exposure is greatest because of their less yielding behavior ( 36 , 50 ). Also, the placement of rumble strips in the right-turn lane can help visually impaired pedestrians to judge whether drivers are yielding as they approach the crosswalk.
Interquartile range (IQR) plot of speed profiles of the slip lane traffic toward the exit crossing point based on their distance from the zebra crossing (left): Illustration of the traffic flow trajectory at the slip lane, (right): IQR plot of the speed distribution for the slip lane traffic.
Conflict Analysis across Slip Lane Designs at Varying Pedestrian Activity Levels
The findings from Figure 9 indicate that the presence of AVs in yield-control slip lanes has different effects on crash severity for AV-pedestrian and HDV-pedestrian conflicts. A decrease in TTC indicates a higher probability of a collision and, according to Gettman and Head, the collision probability is considered as a normalized severity dimension of the safety ( 51 ). As the AV penetration rate increases, the TTC mean value decreases for AV conflicts and increases for HDV conflicts. This suggests that the presence of AVs in yield-control slip lanes leads to less severity in HDV-pedestrian conflicts. On the other hand, for free-flow slip lanes, the TTC mean value slightly decreases for HDV conflicts but shows a drastic decrease for AV conflicts from the 50% to the 75% scenario. This implies that the presence of AVs in free-flow slip lanes increases crash severity for HDV-pedestrian conflicts, while it only increases severity for AV-pedestrian conflicts at higher AV penetration rates. The decline in TTC values as AV penetration rates rise in free-flow slip lanes is likely because of the greater impact of increased speeds characteristic at free-flow slip lanes over the yield-control slip lanes. Therefore, in line with suggestions given by the FHWA’s roundabout information guideline, free-flow slip lanes have a greater negative impact on pedestrian safety than yield-control slip lanes and should not be implemented in areas with higher pedestrian activity ( 22 ).
Comparison of time-to-collision (TTC) mean value for the two slip lane designs: (a) based on human-driven vehicle-pedestrian (HDV-Ped) conflicts and (b) based on autonomous vehicle-pedestrian (AV-Ped) conflicts.
To examine how different slip lane designs influence pedestrian safety at roundabouts across various levels of pedestrian activity, this study considered the entry and exit crossing point for the yield-controlled design and free-flow design and compared them with the normal roundabout traffic approach. The findings (Figure 10) show TTC mean value indicating varying efficacy of slip lane designs in relation to pedestrian traffic demand. At low pedestrian activity (<20 pedestrians/hr). Free-flow lanes show marginally higher TTC values than the yield-control slip lanes at the entry crossing point, but a lower TTC values at the exit crossing point, suggesting a lower severity and probability for crashes at free-flow slip lanes. The presence of the acceleration lane at the exit of the free-flow slip lane drives the conflict severity at the exit. At medium pedestrian activity (20–80 pedestrians/hr), yield-control slip lanes show higher TTC values, which indicate safer interactions between pedestrians and vehicles for both the entry and exit crossing point. At high pedestrian activity (>80 pedestrians/hr) normal approach lanes display the highest TTC, suggesting that traditional roundabout designs without separate slip lanes may be safer in high pedestrian traffic conditions compared with yield-control slip lanes. Based on these insights, the study recommends implementing free-flow slip lanes without acceleration lanes at roundabouts in rural areas with low pedestrian volumes to reduce vehicle delays while maintaining safety ( 46 ). For moderate pedestrian volumes, yield-controlled lanes are advised because of their ability to balance traffic flow and enhance pedestrian safety effectively. In regions with high pedestrian traffic, traditional roundabout approaches are preferable, as they inherently slow down traffic, reduce severe pedestrian conflicts, and elevate safety levels in densely populated or heavily trafficked areas. These recommendations align with FHWA’s roundabout information guidelines, advocating for design adaptations based on specific traffic conditions and pedestrian volumes to optimize safety at roundabouts ( 22 ).
Comparison of time-to-collision (TTC) mean value for the two slip lane designs at varying pedestrian volume: (a) at the entry crossing point of the slip lanes and (b) at the exit crossing point.
Conclusion
The study examined the impact of slip lane designs on pedestrian safety at roundabouts. Slip lanes, which provide a dedicated turning lane for vehicles, have been implemented in various urban areas. However, concerns have been raised about the potential negative effects of slip lanes on pedestrian safety. To investigate this issue, a comprehensive analysis was conducted through simulation modeling and calibration using field trajectory data. The results indicated that slip lane designs significantly influenced pedestrian safety outcomes.
From statistical analysis of the field observed trajectories obtained from a study done by Zhang et al., it is found that a slip lane design can significantly affect the speed of vehicles ( 37 ). It is worth noting (as shown in Figures 8 and 9) that the speed distribution of traffic on the slip lane is considerably higher than that of the approaching traffic at the entry location. This acceleration tendency can be attributed to drivers perceiving the slip lane as an exit from the roundabout. However, this behavior has a great impact on pedestrian safety at roundabouts with slip lanes. Specifically, the design of yield-control slip lanes primarily affects pedestrian safety at the entry crossing point, while the design of free-flow slip lanes affects pedestrian safety at both entry and exit crossing points because of the presence of acceleration lane at exit.
After conducting a simulation analysis using VISSIM and SSAM models, statistical analysis was performed on the SSM indicators, conflicts, and scenarios to examine the impact of AVs on pedestrian safety at slip lane crossings. Results showed that AVs have a positive impact on the occurrence of pedestrian conflicts. The AVs’ TTC values were observed to be much lower than HDVs and decreased further as the AV penetration rate increased. Therefore, this warrants a need to establish more speed control at slip lanes in the future. However, a substantial decrease of 34.62% in the occurrence of pedestrian conflicts was observed as the AV penetration increased to 75%. This significant reduction in pedestrian conflicts with increasing AV penetration rates highlights the potential positive impact of AV technology on enhancing pedestrian safety at roundabouts with slip lanes.
Analysis of the impact of slip lane designs on pedestrian safety was performed. Firstly, the SSM indicators were evaluated based on each slip lane design and vehicle type involved in the conflict. Secondly, a comparison was made of pedestrian conflicts developed between the two models across varying levels of pedestrian activity. From the analysis, it is evident that free-flow slip lanes, while facilitating smoother vehicle flow in low pedestrian traffic areas, present increased risks at higher pedestrian volumes because of accelerated vehicle speeds at exit points. In contrast, yield-controlled slip lanes consistently offer a safer environment for both pedestrians and vehicles by effectively managing vehicle speeds and interactions at entry and exit points. This is particularly true in areas of medium-to-high pedestrian activity, where the controlled flow significantly reduces the likelihood of pedestrian-vehicle conflict. Moreover, traditional roundabout designs without dedicated slip lanes prove to be the safest option in high-pedestrian-traffic conditions. These designs inherently slow down traffic, leading to a substantial reduction in pedestrian conflicts and enhancing safety in densely populated or heavily trafficked areas. The findings underline the importance of tailoring roundabout designs to local conditions, considering both the volume of pedestrian traffic and the proportion of AVs in the traffic mix.
Limitations and Future Research
This study’s analysis was constrained by the lack of pedestrian trajectory data, which necessitated reliance on simulation modeling to evaluate vehicle-pedestrian interactions at pedestrian crossings. Future research should incorporate field studies that capture vehicle-pedestrian interactions, both when pedestrians are present and absent at crossings, to better understand the significant factors that influence drivers’ speed and yielding behavior which affect pedestrian safety within the right-turning slip lane. These significant factors include, but are not limited to, curb radius, channelizing island, and striped pathway ( 52 ). Further analysis of this topic would provide adjustments to the current guidelines on right-turning slip lanes ( 45 ).
Recommendations
In light of the findings about the impact of slip lanes on pedestrian safety at roundabouts, particularly considering the integration of AVs, recommendations have been suggested. Firstly, optimizing slip lane design by adjusting curvature or alignment can ensure clear lines of sight for pedestrians and vehicles. This can be achieved by implementing a compound curve radius with a larger radius preceding a tighter curve to allow for gradual deceleration of vehicles ( 46 ). Additionally, implementing traffic calming measures, such as advanced yield marking, enforcing speed limits near roundabouts with slip lanes, and avoiding slip lanes with larger radii, are essential to mitigate the influence of slip lanes on driver speed. Moreover, mandating the implementation of signalized pedestrian crossings at slip lanes, along with the inclusion of pedestrian delay models for the accommodation of pedestrians with disabilities, is crucial for enhancing pedestrian safety ( 53 , 54 ). Also, implementing recent information-sharing and local collaborative learning methods can enhance AV awareness and responsiveness, reducing the likelihood of conflicts with pedestrians. Finally, the installation of raised crosswalks at roundabout entries can significantly improve pedestrian safety and visibility, particularly in areas with high pedestrian activity ( 13 ). These recommendations collectively aim to enhance pedestrian safety and mobility in the context of evolving transportation landscapes.
Footnotes
Acknowledgements
The authors sincerely acknowledge the valuable contribution of Zhang et al.’s study (37), which provided the vehicle trajectory data captured by their roadside perception system. This dataset, supported by the University of Michigan’s mobility center and test facility (Mcity) and the National Science Foundation (NSF), has played a pivotal role in our research, significantly enriching the depth and quality of our findings.
Author Contributions
The authors confirm their contribution to the paper as follows: study conception and design: V. Kwigizile, N. Novat; data collection: N. Novat; analysis and interpretation of results: N. Novat, V. Kwigizile, J. Oh; draft manuscript preparation: N. Novat, V. Kwigizile, J. Oh. All authors reviewed the results and approved the final version of the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
