Modeling Behavior of U-Turning Vehicles at the Median Opening Using a Merging Behavior Approach: A Case Study in Bahir Dar City, Ethiopia

Median openings are one of the most commonly used road features, which are mainly used to allow U-turning movement in urban areas, and this study focuses mainly on modeling the behavior of U-turning vehicles at the median opening using a merging behavior approach. The purpose of the study is to estimate and model the critical gap of u-turning vehicles at the median opening under mixed traffic conditions. Under this study, the accepted gap, rejected gap, driver waiting time, merging time, and critical gap are estimated, and the modified Raff's method and modified INAFOGA method are used for the estimation of a critical gap. However, modified INAFOGA is used for the modeling of critical gaps under mixed traffic conditions. In this study, sixteen median openings were selected in Bahir Dar city, and data were collected using a video recording technique at each selected median opening during the peak hour of the day. The necessary data were extracted using Forevid analysis software tools. Different types of traffic are involved in the mixed traffic, and each vehicle type is categorized according to the Ethiopian Road Authority's 2013 design guide into seven different classes, such as 2-wheeler, 3-wheeler, passenger car, minibus, small bus and truck, medium bus, and medium truck. Among those traffic types, three vehicle classes (three-wheeler, passenger car, and minibus) were only considered due to the prohibition of U-turning movement for medium and large vehicles. For the modeling of critical gaps, waiting time and conflicting traffic flow are used as independent variables using the regression technique. Driver waiting time and the critical gap were found to be power related to passenger cars and minibuses and exponentially to three-wheelers. Conflicting traffic flow and critical gaps were power related to passenger cars and minibuses and linearly related to three-wheelers.


Introduction
Median openings are one of the most common road facilities used to facilitate the j-turn and U-turn movement of vehicles in Ethiopia, and they are preferred to overpasses and other types of road facilities due to their low cost. Since Bahir Dar is a city located in Ethiopia, median openings are the most common road facilities to facilitate U-turn movement of motorized and nonmotorized vehicles. Te majority of signalized intersections do not allow U-turns since doing so would maximize the intersection's confict spots and enhance its service quality. Te most frequently used road facility for U-turn vehicles is a median opening in Bahir Dar due to the city's heterogeneous trafc, which includes a wide variety of vehicle classes. To give U-turn vehicles enough room for safe merging, the through trafc stream is forced to slow down, while U-turn vehicles attempt to enter the median opening. Te spaces supplied to U-turning cars are not the normal time headway due to these trafc phenomena, which have an impact on the behavior of the through trafc stream. Te entry capacity of the lower priority stream (U-turning vehicles) is signifcantly impacted by this situation, which also delays the higher priority trafc stream and results in crashes between lower and higher priority trafc streams. Tis situation forces the acceptance of the gap to be governed by priority rather than natural trafc fow or adherence. Due to all of these factors, median openings are a crash-and delay-prone region. U-turn vehicles at median openings need to be examined by the notion of the parameters which afect the U-turn vehicles at median openings due to the unpredictability of the gapaccepting behavior of U-turn vehicles at median openings.
Terefore, the median opening of separated urban highways is provided to increase intersection efciency by reducing confict points, and those openings help provide access to opposing trafc quite well without interfering with priority trafc or U-turn vehicles. To make a safe turn, a U-turning vehicle will always wait for a suitable gap in the subsequent priority trafc stream [1]. Another researcher [2] briefy describes the U-turning behavior of vehicles at the median opening, and the author uses the sectioning method for the study of their behavior and discusses that vehicles enter the U-turn roadway with steadily reduced speed and search the priority trafc stream for a safe exit merging gap at the entry section. On arrival at the exit section, drivers make their intentions known to the priority stream by moving, sometimes menacingly, to the upper end of the exit lane. Priority vehicles may allow the merge to occur, increase speed, change lanes to avoid collusion, fash their headlight, or simply ignore the nonpriority vehicle altogether. At the peak period, merging from the exit point becomes a difcult maneuver and a daring afair.
Due to the high speed, high trafc congestion, and the requirement that the turning vehicle undertake a reverse movement, U-turn movements in median openings are more complicated and dangerous than conventional turning movements at signalized and unsignalized intersections [3]. Te turning vehicle must wait and then turn under lowspeed conditions in the face of oncoming trafc, which means conficting trafc may need to accelerate rapidly to achieve the speed of the trafc stream. To analyze the behavior of U-turning movements at the median opening, estimating the critical gap is the most important thing, and as per [4], the critical gap can be defned as the minimum time interval between the through-trafc stream vehicles that is necessary for U-turning vehicles to make a merging maneuver. Gupta et al. [1] state that the values of the critical gap that are accepted vary for diferent vehicle classes and depend on various parameters, such as the type of U-turning vehicles, several stream parameters of opposing lane trafc, and the geometric elements of the carriageway, including the median.
Diferent methods have been used by diferent researchers for a while. For heterogeneous trafc fow conditions, the authors in [5] used existing methods like probit, Hewitt, modifed Raf, logit, and Harder methods for the estimation of a critical gap at unsignalized intersections. Tere was a wide diference (12%-38%) between the critical gap values, which highlighted the limitations of the methods to address mixed trafc situations. Tus, the authors came up with an alternative technique that makes use of the clearing behavior of the driver in conjunction with gap acceptance data. Te new method developed in this study was simple and easy to implement under Indian conditions. With this in mind, Pannela and Bhuyan improved the clearing behavior assumed for the unsignalized intersection to the merging behavior of U-turn vehicles at the median opening [6]. Tis method takes a vehicle's gap-accepting characteristics into account in addition to its actual merging behavior. Assuming merging time shows how a movement is carried out at the median opening. As an alternative, it considers the challenges presented by mixed-trafc conditions, including lane discipline and the rule of priority impact.
Tis paper aims to estimate and model the critical gap of a U-turning vehicle at the median opening and estimate the critical gap using the merging behavior approach. In addition to estimating the critical gap, we consider the efect of the driver waiting time and volume of conficting trafc fow on the critical gap of U-turn vehicles.
Te paper makes two basic contributions. Te frst is initiating the study of U-turning vehicles at median openings since the study is new in Ethiopia. Tis will open the door for other potential researchers to investigate and study more on the U-turning behavior and its impact on trafc streams. Te second is estimating and modeling the critical gap of U-turning vehicles under mixed trafc conditions, which will help the transportation agencies fully understand the behavior of the trafc fow at the median opening under U-turn movements.
Related works in diferent literature studies are discussed in Section 2. Te data collection for the estimation of the critical gap, the data extraction as per the merging behavior approach, and the procedure used for the estimation and modeling of the critical gap are discussed in Section 3. Indepth results and discussion obtained from the estimation and modeling of the critical gap are mentioned in Section 4. Te conclusions obtained from the detailed results and discussions are mentioned in Section 5, and recommendations based on the fndings are discussed in Section 6. Finally, the limitations and gaps of the study are discussed in Section 7.

Related Works
In this section, we briefy discussed the related work, which focuses on the behavior of U-turning vehicles at the median opening and the estimation and modeling of the critical gap for U-turning vehicles.

Behavior of U-Turning Vehicle Movement.
Te gap acceptance behavior of U-turn vehicle drivers is highly afected by the waiting time, and the statistical result from the study by Shubber showed that when the driver's waiting time fell in a range between 21 and 30 sec, the driver was forced to accept a gap size less than that, which fell in the range of 11 to 20 sec at a confdence interval of 95%. On the other hand, there is a slight diference in the mean gap acceptance between an interval of (1-10) and (11)(12)(13)(14)(15)(16)(17)(18)(19)(20) sec at the same confdence interval [7]. In addition to the drivers' waiting time, another researcher [8] demonstrated that gender has an impact. It was shown that male drivers tend to be more aggressive and take greater risks when doing U-turns. Likewise, commercial drivers were found to be more aggressive than personal vehicles and were found to complete the U-turn using less width of the carriageway in the opposing lane compared to personal vehicles. Furthermore, loaded vehicles are more careful and turn their steering slowly compared to empty vehicles while taking U-turns.
A U-turn at a midblock median opening is a complicated maneuver that causes drivers to be unsure whether to accept or reject the available gap in order to prevent a collision with opposing trafc fow. Te study by Khan and Mohapatra in 2022 analyzed the dilemma zone for U-turning vehicles at a midblock median opening for diferent vehicle categories, and they concluded that the dilemma zone would be highly afected by the vehicle size [9].
Te study by Khan et al. [10], which focused on estimating the temporal and spatial critical gap of U-turning vehicles at uncontrolled median openings for six diferent types of U-turning vehicles, revealed that the critical gap values discovered during the research are signifcantly smaller than those reported in developed countries. Tis fnding illustrates the aggressive driving nature of drivers in developing countries. In 2022, Mazaheri et al. [11] state that the critical gap for drivers of heavy vehicles was nearly lower than that for drivers of cars, indicating that heavy vehicle drivers were acting more aggressively, and another study, which is conducted on the aggressive behavior of drivers at an uncontrolled intersection under mixed trafc conditions, showed that with an increase in the lag/gap value and a decrease in clearing time, the probability of accepting a gap increases. Te likelihood also reduces as major road vehicle size grows, whereas minor road vehicle size increases. Te lag/gap and speed of major road vehicles decrease, so do the aggression levels of small road vehicles, while they decline as minor road vehicle size and clearing time grow [12].

Development of the Critical Gap Model.
Te critical gap of a U-turning vehicle at the median opening is highly affected by diferent parameters. In 2011, El Esawey and Sayed [13] conducted a study on the operational performance of U-turns at the median opening, and they came to the conclusion that the amount of trafc on the major stream has a substantial impact on the outcome of the U-turning maneuver. Tey also stated that the median openings function well in light to moderate trafc, unless the volume of the competing trafc efect increases to the point where it becomes impossible to make a U-turn.
Te merging behavior approach is the most preferable approach to account for the efect of mixed trafc conditions and lane discipline efects on the gap acceptance behavior of drivers, and the critical gap has a strong correlation with the gap between conficting trafc, the volume of conficting trafc, and the conficting trafc speed [4].
As per Datta, critical gaps are strongly correlated with conficting trafc volume, conficting trafc speed, and U-turn vehicle waiting times. Te model prepared by Datta for four vehicle classes (2W, 3W, 4W, and SUV/MUV) showed that the critical gap has a strong linear correlation with conficting trafc speed and a strong power correlation with waiting time for each vehicle class and also a strong power relationship with conficting trafc volume for each vehicle class except for 2W, which has a linear correlation. In 2017, Dash et al. [14] developed a model that correlates the critical gap with the conficting trafc volume and proved that conficting trafc has a signifcant linear relationship with the critical gap of U-turning vehicles.

Research Methodology
To analyze the behavior of U-turn vehicles at the median opening, the estimation of the critical gap is the key to a successful design and analysis. However, it is not possible to measure the critical gap directly from the feld due to its complexity. Even if it is not possible to measure the critical gap directly from the feld, some important parameters can be measured there and help estimate the critical gap. Since accepted gaps and rejected gaps can be measured in the feld, we can use them to estimate the critical gaps using diferent methods. In this study, two approaches, modifed Raf's and infuence Area for Gap Acceptance (INAFOGA) methods, were used for the estimation of a critical gap, and the procedure used for the estimation and modeling of a critical gap is presented in Figure 1.

Data Collection.
Data collection was performed with the help of video recording techniques at each of the sixteen selected median openings. Recording was performed by fxing the camera on a tripod stand at the top of the building so that we could capture the entire median and its vehicular movement. Te data were collected during the peak and ofpeak hours of weekdays. Due to the signifcant variance in data sets, weekend days and public holidays are typically ignored, resulting in inaccurate estimation of important gaps in U-turning trafc near median openings [15]. Te data collected through video recording techniques are shown in Figure 2.

Data Extraction as per the Merging Behavior Approach.
Te video data were analyzed frame by frame with an accuracy of 0.03 s based on a frame rate of 30 frames per second using Forevid analysis software to extract and determine all necessary parameters like merging time, gap acceptance, waiting time, rejected gap, and conficting trafc volume. Video recording of all sixteen median openings and through vehicles is counted. Trough vehicle trafc comprised vehicles including two-wheelers, three-wheelers, passenger cars (pcs), minibuses, small buses, medium buses, small trucks, and medium trucks, and their respective passenger car units are shown in Table 1as classifed by the Ethiopian Road Authority [16]. Te classes of U-turning vehicles considered in this study are three-wheelers, passenger cars with a seat of fewer than ten passengers, and minibuses with seats between 11 and 14.
Computational Intelligence and Neuroscience Tose vehicle classes are considered for this study's purpose as a U-turning vehicle at each selected median opening due to the prohibition of U-turn movement at the median opening by the city municipality; medium and large vehicle classes are not considered as U-turning vehicles.
Te merging behavior approach is the most preferable method for the estimation of the critical gap when the trafc condition is mixed, lane discipline is not enforced, and the rule of priority is not obeyed or practiced [3,6,17,18]. Te frst step of the merging behavior approach is to identify the infuence area for gap acceptance (INAFOGA), which is an area, where it is believed that all merging and gap acceptances occur.

Determination of the Infuence Area for Gap Acceptance (INAFOGA).
After the video data are recorded and the necessary data related to the geometric future of the median opening are collected, data are extracted to get the diferent necessary parameters for the estimation and development of the critical gap model. Te frst thing we need to do for the modeling of the critical gap is identify the INAFOGA.
As per Mohapatra et al. [19], the identifcation of the critical position INAFOGA is afected by the size of the vehicle, and it is shown that a smaller vehicle has a large standard deviation compared to larger vehicles on the uniqueness of the INAFOGA. In the identifcation of INAFOGA, diferent researchers have recommended different methodologies, but as per [3,19], the authors use the merging behavior approach for the identifcation. In this approach, the cumulative clearing time was calculated as the time required by a lower-priority vehicle to clear the INAFOGA, the curve was plotted, and the cumulative frequency of the corresponding lag and gap acceptance was calculated for each vehicle category after the intersection of the curves, which is considered the critical gap. Te identifcation of INAFOGA is performed, as shown in Figure 3.
After a careful look at the video data collected at each specifc median opening, the position of each line of the INAFOGA is determined as follows:    However, the authors in [3] state that the identifcation of upstream boundary BC and lower boundary CD will have less signifcance in estimating the merging time of the U-turning vehicle.
After the infuence area for gap acceptance (INAFOGA) is identifed, the basic data for the estimation of the critical gap using the modifed INAFOGA and modifed Raf's methods are extracted.
Te gap is estimated as the time diference between the arrival of the consecutive through vehicle and line BC. Te merging time is the time taken to traverse the INAFOGA without causing any crashes with conficting trafc, and it is measured from the time the rear bumper passed over stop line AB to the time the rear bumper crossed merging line AD. Driver waiting time is the time the U-turning vehicle spends waiting for the gap sufcient for full merging after he/she reaches the stop line of the INAFOGA.

Estimation of Critical Gaps.
Te critical gap (t c ) is the minimum time interval between the through-trafc stream vehicles that is necessary for a U-turning vehicle to make a merging maneuver. Te value of the critical gap can difer, depending on the driver's behavior, which is why some drivers accept a smaller gap and others accept a too long gap based on the driver's behavior. Diferent methods have been used throughout the year for the estimation of the critical gap since it cannot be measured directly in the feld. However, there are many diferent methods used for the estimation of critical gaps. Among those methods, the modifed INAFOGA method is the one that is designed to address the mixed trafc condition under lane discipline and the rule of priority efects, and due to this, the method is used for the modeling of the critical gap. Te modifed Raf's method is used for the comparison of the model results in the validation process.

Modifed Raf's Method.
Te earliest method for estimating the critical gap was frst proposed by Brilon et al. [20], and the authors defned that the critical gap t c is the value of time t which is a function of 1 − F r (t) and F a . (1) Tis method involves the empirical distribution functions of accepted gap F a (t) and rejected gap F r (t). As per the Raf method, a critical gap at unsignalized intersections is defned as "the gap/lag for which no. of accepted gaps shorter than it is equal to the no. of rejected gaps longer than it." Te arrival of mainstream vehicles can be described by a Poisson distribution but only for light-medium trafc fow conditions. Raf's method involves the extraction of the length of the gaps in seconds, for which the driver waits at the median opening to accept a suitable gap, an accepted gap, or a rejected gap. Two cumulative distribution curves will be drawn with no gaps in the ordinate and gap size as abscissa, and these two cumulative distribution curves will relate gap length (t) with the accepted gap and rejected gap. After the curves are drawn, the intersection of the two cumulative curves will be taken as a critical gap for that specifc vehicle type [21][22][23].

Modifed INAFOGA Method.
Te new method developed in this study was simple and easy to implement under Indian conditions. With due consideration, this paper has provided signifcant background for the present study because of its heftiness towards mixed trafc conditions prevailing in Ethiopia. Te "clearing behavior" assumed for unsignalized intersections in the previous study was improved to the merging behavior in the case of U-turn vehicles at median openings in this study, as [14, 17, 24, and 25] used in their study. It considers the actual merging behavior in addition to the gap acceptance features of a vehicle. Merging time shows how the movement is implemented at the median opening. It also takes into account the difculties found under mixed trafc conditions.
In this method, a cumulative distribution graph that contains the gap size on the x-axis and the cumulative percentage of gaps on the y-axis and two cumulative percentage graphs of the merging time and accepted gap will be prepared, and the intersection of the two will be taken as the critical gap of the specifc vehicle class.

Modeling of the Critical Gap.
Modeling of the critical gap is performed as per the merging behavior approach, and the critical gap estimated using the modifed INAFOGA method was used as the dependent variable. Independent variables are frst selected based on the efect they have on the estimated critical gap value as per diferent research, and researchers have identifed conficting trafc fow, conficting trafc speed, width and number of the lanes, the geometric feature of the median opening, and drivers' waiting time as having a major impact on the estimation of the critical gap value and mainly afecting the driver's decision. Among those parameters, conficting trafc fow and waiting time were considered the major parameters for the modeling of the critical gap.
Since the critical gap is estimated based on the merging behavior approach, the merging time and accepted gap are the variables used for the estimation of the critical gap.

Critical Gap Estimation
(a) Modifed Raf's method: the critical gap for all data points in each median is estimated as the number of accepted lags shorter than the critical lag and is the same as the number of accepted lags longer than the critical lag [23]. Te critical gap for each U-turning vehicle (3W, pc, and mb) is estimated, as shown in Figure 4. (b) Modifed INAFOGA method: after the merging time and accepted gap data are extracted from the video recorded according to the merging behavior approach, the cumulative frequency distribution curve of the merging time and accepted gap will be plotted, and the point of the intersection of the two curves at the x-axis will be taken as the critical gap of the specifc vehicle class. Te result of the merging modifed INAFOGA is shown in Figure 5.
Te summarized result of the critical gap estimation using both methods is presented in Table 2 for each of the three U-turning vehicle classes at each of the sixteen median openings.

Modeling of the Critical Gap.
Te critical gap of U-turning vehicles at the median opening is modeled in terms of the driver waiting time and conficting trafc fow. Driver waiting time is the time that the U-turning vehicle spent waiting for the gap sufcient for full merging after he/she reached the stop line of the INAFOGA, and it will be extracted from the video recording data from each selected median opening. Te conficting trafc fow is counted from the video recorded, and the hourly trafc volume is converted using their respective passenger car units (pcus) as recommended by the Ethiopian Road Authority [16]. Seven diferent types of vehicle classes were involved in our study, as shown in Table 1.
Te correlation for the development of the regression model was performed between the critical gap and waiting time and the critical gap and conficting trafc fow for the U-turning vehicle class.

Critical Gap vs Driver Waiting Time Model. It has been
shown in diferent research studies that waiting time varies from driver to driver based on their age, sex, and other behaviors according to the gap they accept [26]. Diferent studies proved that the waiting time afects the critical gap at unsignalized intersections as well as at median openings, and as the waiting time increases, the critical gap becomes smaller, which is due to the driver's loss of patience for waiting and accepting the smaller gap than the one they rejected before. Te waiting time for the modeling of the critical gap is taken as the average waiting time of every hour for modeling.
(i) Tree-wheeler (3W) vehicle: in the regression model, the exponential curve ftting shows a good Rsquare ft to the relationship between the critical gap and drivers' waiting time, and in this relationship, as the drivers' waiting time increases, drivers lose patience and tend to accept smaller gaps. Te variation of the critical gap with driver waiting time is shown in Figure 6.
(ii) Passenger car (pc): the curve-ftting result has shown that for the vehicle category passenger car, the correlation between the critical gap and the driver waiting time is a power relationship, as shown in Figure 7.
(iii) Minibus (mb): the model result showed that the critical gap and driver waiting time is power related to a good ft R-square, as shown in Figure 8.

Critical Gap vs Conficting Trafc Flow.
Sometimes the U-turning maneuver may be taken in a jam condition without any sufcient gap [27,28], and this sometimes leads to collisions. Diferent research studies have proved that conficting trafc fow has a signifcant efect on the estimation of the critical gap, and due to this, diferent models have been prepared by so many researchers in the past that show the correlation between conficting trafc fow and the critical gap [19,29,30].
(i) Tree-wheeler (3W): the critical gap of a U-turning 3W vehicle is correlated with the average hourly conficting trafc fow and their relation ftted into some mathematical models. Te model showed that the critical gap and conficting trafc fow are linearly related, as indicated by the best ft R-square. Te variation of the critical gap with conficting trafc fow is shown in Figure 9.
(ii) Passenger car (pc): unlike three-wheeler vehicle classes, the relation between the conficting trafc fow and the critical gap has a power relationship, and as the conficting trafc fow increases, the critical gap decreases due to the small size gap provided for U-turning vehicles. Te relationship between the critical gap and conficting trafc fow is shown in Figure 10.
(iii) Minibus (mb): the correlation between the conficting trafc fow and critical gap for a minibus is best ftted through the power relationship of the model, and as the conficting trafc fow increases, the critical gap value reduces. Te model is best ftted with an R-square value of 0.61, as shown in Figure 11.       Computational Intelligence and Neuroscience       Tables 3-8, the null hypothesis will be accepted.

Conclusion
Te present study estimates the critical gap using both modifed Raf's and modifed INAFOGA methods, and the merging behavior approach is used as the base approach for the modeling of the critical gap since the approach can account for the impact of mixed trafc conditions, violations of rules of priority, and lane discipline. Te modifed Raf's method result showed that the maximum and minimum critical gaps for three-wheelers are 4.4 and 3.3 seconds, 6.38 and 4 seconds for passenger cars, and 5.82 and 4.37 for minibuses, respectively. Te result showed that threewheelers take smaller available critical gaps due to their smaller size, while passenger cars take the higher critical gap due to a lack of driving experience in comparison with minibuses. Since most passenger cars are used for private use, minibuses are used as a taxi mode of the transport system, andas a result of this, passenger car drivers tend to accept larger gap due to their lack of driving experience. In the modifed INAFOGA method, the maximum and minimum critical gaps for three-wheelers are 4.77 and 2.77 seconds, for passenger cars, they are 5.51 and 3.43 seconds, and for minibuses, they are 7.01 and 3.49 seconds, respectively. For critical gap modeling, two independent variables (drivers' waiting time and the conficting trafc fow) were taken into account. Te critical gap and those independent variables, whose probability value is less than the signifcance level (5%), were empirically related, using the regression technique for each U-turning vehicle, and the model result showed a high coefcient of determination (R 2 ). From the model result, conficting trafc fow has an inverse relationship with the critical gap, and as the volume of conficting trafc fow increases, U-turning drivers accept smaller gaps. Te drivers' waiting time has an exponential correlation with the critical gap and a linear correlation with the conficting trafc fow for three-wheeler vehicles. For passenger cars and minibuses, both drivers' waiting time and conficting trafc fow showed a power relationship with the critical gap.
It is important to note that there are many factors to be considered in the study of the U-turn behavior of the vehicle,   such as the geometric feature of the median opening, the gender of the driver, and the trafc management system. Further research needs to be performed on the efect of those factors on the critical gap of U-turning vehicles at the median opening.

Recommendations
Te result of this study may be used for the analysis and modeling of the capacity of median openings so that U-turning trafc movement becomes safe and efcient. In addition to this, the study can be used in the study of the efect of U-turning vehicles on the fow of the major trafc stream and in delay and queue analysis.

Limitations
Tis study looks into the gap acceptance behavior of U-turning vehicles at the median opening and the efect of conficting trafc fow and drivers' waiting time on the critical gap. However, even if the critical gap is highly affected by these parameters, it is also afected by the geometric feature of the road width and conficting trafc speed. It is certain that the speed of conficting trafc speed and geometric features of the road highly infuence the critical gap of U-turning vehicles, which is proposed to be investigated in a future study. Te efect of driver experience, gender, and age is also proposed to be studied in feature work.

MO:
Median opening t: Te time headway between successive vehicles in the major stream (second) t c : Critical gap (second) INAFOGA: Infuence Area for Gap Acceptance 2W: Two-wheeler 3W: Tree-wheeler 4W: Four-wheeler pc: Passenger car SUV: Sport utility vehicle MUV: Multiutility vehicle mb: Minibus.

Data Availability
Te data that support the fndings of this study are available from the corresponding author upon request.