Empirical Study on the Development of Driving Environments for Personal Mobility Vehicles

Abstract

Personal mobility vehicles (PMVs) are the potential future travel mode, typically for “first-mile and last-mile trips,” in urban areas. Accordingly, the popularity of PMVs is increasing, and the purpose of using PMVs is diversifying in South Korea. Before revision of the law concerning PMVs, the driving ways and roads for PMVs were not defined by regulations in South Korea. Therefore, most PMV users preferred to drive on sidewalks and bicycle lanes to avoid collisions with larger vehicles. Furthermore, PMVs have been involved in many accidents. To solve these problems, the government of South Korea revised the Road Traffic Act to permit PMVs to travel on bicycle lanes and considered building a new type of lane for PMVs based on the results of field experiments conducted in this study. In these experiments, the minimum turning radius with various speeds, driving performance on roads with vertical alignments, lateral placements in curved sections were examined. In addition, conflicts and inconvenience of PMVs driving on bicycle lanes were tested. In the experiments, three different types of PMVs—electric scooter (e-scooter), electric unicycle (e-unicycle), and segway—were used. Through the experiments, the feasibility of using PMVs on bicycle lanes was assessed, and the design criteria for PMV lanes were derived.

Keywords

personal mobility vehicle bicycle lanes driving performance field experiment design criteria

A personal mobility vehicle (PMV) could be defined as a single-person mode of transportation powered by electricity, a definition we have derived from the review of related literature. The classification criteria for PMVs have not been established, and therefore, the concept includes many types of devices. PMVs can be an alternative mode of transport suitable for short-range travel, especially for “first-mile and last-mile trips” in an urban area. Recently, the number of PMV users has steadily increased in South Korea, so the need for adequate road environment and institutions associated with PMVs has risen as well. To meet the demand, the South Korean Road Traffic Act was amended to include the road types allowed for driving PMVs and the transports that can be classified as PMVs. Before the amendments, PMVs could be driven only on a roadway, but users preferred to drive on sidewalks and bicycle lanes to avoid collision with larger vehicles. Furthermore, there have been many accidents, typically between cars and PMVs, as well as between pedestrians and PMVs. The South Korean government was aware of these troubles; thus, the law was revised to allow PMVs to be driven on bicycle lanes. The government’s decision was made based on the problems with driving PMVs on bicycle lanes assessed in this study. Furthermore, the Korean government considered building a new type of lane only for PMVs as an alternative for PMV users.

This study was performed to investigate the availability of bicycle lanes for PMVs through field experiments by analyzing the various driving behaviors of PMVs to permit them to be driven on bicycle lanes. Furthermore, in this study, assuming that road space is allocated for PMVs, the road design standards for PMV lanes, which are necessary for making a safe and convenient place in the road environment for PMVs, were acquired.

Literature Review

Definition of PMVs

Each research group has defined PMV differently. Kim et al. ( 1 ) defined PMV as a “personal low-speed means of transportation using electricity as the main/auxiliary power source.” Han et al. ( 2 ) defined PMV as a “new concept of electricity-powered transportation for one person.” On the other hand, there have been studies that defined PMV as “a means of transportation mainly powered by electricity for one or two persons” based on the number of people available for boarding and the power source ( 3 – 6 ). In addition to the number of passengers and power sources, there were also studies that defined PMV driving distance. Woo ( 7 ) defined PMV as “a means of transportation for one or two persons that can be conveniently used by commuters with short-distance or people with periodic trip pattern,” and Park et al. ( 8 ) defined PMV as “a means of transportation mainly powered by electricity for short and medium distances.” In this study, PMV was defined based on the number of passengers and power sources by referring to the definition of PMV in the related literature. However, because this research was conducted on PMVs that only one person could board, PMV was defined as “a single-person mode of transportation powered by electricity.”

Characteristics and Trends of PMV-Related Accidents

Our research team confirmed that accidents related to PMV have steadily increased in recent years, through the traffic accident data collected by the Korea Road Traffic Authority (KoROAD). According to accident statistics collected by KoROAD, the number of PMV-related accidents increased by 96% annually from 117 in 2017 to 1,735 in 2021. In addition, the number of injuries in PMV-related accidents steadily increased, and the number of injuries increased by 98% annually from 124 in 2017 to 1,901 in 2021. Accordingly, the number of deaths from PMV-related accidents also increased by 54% annually, from four in 2017 to 19 in 2021. As confirmed by these accident statistics, accidents related to PMV have annually increased ( 9 – 13 ). These accident statistics also indicate that current roads do not provide sufficient environmental conditions for the safe driving of PMVs. In addition to the statistics on PMV-related accidents, several studies were conducted on the characteristics of PMV-related accidents and the risk of injury. Lee et al. ( 14 ) and Kim ( 15 ) used a simulation program to study the degree of injury to drivers and pedestrians when a PMV collides with bicycles and pedestrians. They modeled PMVs, bicycles, and drivers for each transportation mode, as well as pedestrians, in the simulation program. As a result of the PMV and bicycle collision simulations, bicycle riders were more likely to be injured than PMV drivers. When an e-scooter and a bicycle collided, the drivers of both means of transportation were less likely to be injured. In contrast, bicycle riders were more likely to be injured when colliding with an e-unicycle or hoverboard. In addition, results from PMV and pedestrian collision simulations showed there was a high possibility of injury to pedestrians. In particular, pedestrians were more likely to hit their heads on the ground while falling after colliding with a PMV. The researchers also analyzed the degree of injury to the head in the event of a collision using the Head Injury Criteria (HIC). When the HIC exceeds 1,000, it is judged that a person has severe head injuries. As a result of the analysis, the HIC exceeded 1,000 five out of 10 times (50%) in the simulation experiment when a PMV collided with a bicycle and 12 out of 14 times (86%) when colliding with a pedestrian. Through these results, it was concluded that driving on the sidewalk could pose a greater risk than riding on the bicycle lanes. Kühn and Grabolle ( 16 ) conducted an experiment to assess the risk of collisions between segways and pedestrians and vehicles. They conducted a crash experiment using dummies, a segway, and a passenger car in which the segway with a driver dummy approached a pedestrian dummy and the passenger car, in turn, at 15 km/h. In the experimental collision between the segway and a pedestrian, the segway and pedestrian were set to collide head-on. As a result of the collision, both the segway driver dummy and the pedestrian dummy suffered severe head injuries. The pedestrian dummy and the segway driver dummy struck the ground as they fell after colliding; both experienced significant impacts on the head, neck, chest. The pedestrian dummy also had significant impact to the pelvis and suffered serious injuries to the leg and ankle as the segway collided with the pedestrian’s leg. In the experiment in which the segway and the passenger car collided, the segway was set to collide with the side of the car. As a result of the collision, it was found that a strong force was applied to the neck of the segway driver dummy, however, the collision did not cause serious injury to the dummy. Although the crash impact did not exceed the condition that could injure the segway driver dummy’s neck, the researchers concluded that segway drivers have the potential to be seriously injured in a collision with a car ( 16 ). Han et al. ( 17 ) analyzed PMV traffic accidents that occurred between 2017 and 2019 in South Korea. The results of the analysis showed that accidents between PMVs and pedestrians occurred more frequently than accidents between PMVs and bicycles. In addition, according to the results of analysis of fatal accidents, 50% of the fatal accidents involving PMV drivers were vehicle-only accidents, and seven out of 10 vehicle-only accidents occurred at night.

Our research team reviewed related research and confirmed that PMV-related accidents have steadily increased in South Korea. Moreover, it was also confirmed that casualties might occur if PMV is involved in an accident. Therefore, we conducted this study to help with policies to provide a safer driving environment for PMV users.

Driving Ways of PMVs

In South Korea, there are some institutional limitations for PMV use. Only people who have a driver’s license are allowed to drive a PMV, and they must wear fully covered helmets for protection against serious injuries. Further, before the revision of the regulations, PMVs were allowed to be driven only on roadways. However, most PMV users did not follow the regulations, and there were many accidents involving PMVs ( 1 , 2 ). Based on these issues, numerous studies raised the need for revision of the regulations related to PMVs (3 –5, 18 , 19 ). Therefore, the law concerning PMVs was revised to permit them to travel on bicycle lanes in South Korea. Before the revision of the law, the regulations related to bicycle lane facilities did not include contents related to PMVs ( 20 , 21 ). However, since the amendment, provisions related to PMVs have been included in the regulations ( 22 ). The permitted pathways for PMVs are different in different countries ( 23 ), as listed in Table 1. Generally, PMVs are permitted to be driven on bicycle lanes and are often limited on car lanes and pedestrian paths.

Table 1.

Permitted Pathways for Personal Mobility Vehicles in Different Countries

Country	Ways
Country	Car lanes	Bicycle lanes	Pedestrian path	Private land
United States
California	○	○	○	○
New York	○	○	○	○
Washington	○	○	△	○
Germany	○	○	×	○
Netherlands	○	○	△	○
Switzerland	○	○	△	○
France	×	○	○	○
Australia (Queensland)	×	○	○	○
Singapore	×	○	×	○
South Korea	○	○	×	○

Note: ○ = allowed to use; △ = conditionally allowed to use; × = not allowed.

Driving Performance of PMVs

To date, several studies on the driving performance of PMVs have been reported. The majority of them were conducted through surveys and analysis of recorded video to investigate the usage of PMVs. The U.S. Federal Highway Administration (FHWA) also conducted experiments to identify any problem that may arise when allowing emerging vehicles to drive on bicycle pathways. The experiments included many types of devices, and some PMVs were also included. For the experiments, 21 data collection stations were installed across the United States. At each station, data from seven different experiments were collected, including physical dimensions, space requirements for turns, sweep widths, turning radii, acceleration capability, speeds, and stopping sight distance (SSD). Based on the experiments, it was concluded that the design guidelines for bicycle facilities may be inadequate for emerging vehicles and it was recommended that they should be revised ( 24 ). Xu et al. ( 25 ) developed risk indicators using the overtaking behaviors by electric and regular bicycles in bicycle lanes and analyzed the variables affecting the indicators. Because of the difficulty of observing overtaking behaviors in the field, two surrogate variables for the overtaking rates—standard deviation and the average speed difference between two neighboring bicycles—were used as the risk factors related to speed dispersions. Further, two different models were developed using some variables, such as the number of lines, traffic conditions, type of bicycles, and type of drivers. Based on the analysis results, it was found that overtaking events were positively correlated with speed dispersions, and the traffic conditions greatly affected the risk indicators when the volume-to-capacity ratio was high. In addition, through regression analysis, it was found that the proportion of electric bicycles had a significant effect on the risk indicators. Li and Ando ( 26 ) conducted a survey of attitudes toward two-wheel vehicles. It was found that those who had driven two-wheel vehicles had a positive attitude toward them. Moreover, “tourism and expansion,”“short-distance trips in downtown areas,” and “traversing within buildings” were the main purposes of using the devices. On the other hand, “last-mile” and “business travel in urban areas” declined significantly after using them. Kim et al. ( 1 ) investigated patterns of riding bicycles and PMVs on the roadway and wearing a helmet, based on survey results. According to the survey, PMV users traveled shorter distances than bicycle users, and PMV users drove along pedestrian paths more frequently than bicycle users. In addition, the rate of wearing a helmet was significantly lower for PMV users than for bicycle users. Woo ( 7 ) devised a standard test for evaluating driving performance of PMVs, suggesting two tests: one to measure the maximum speed and the other to evaluate it on a slope.

In addition, studies for assessing the driving performances of PMVs were conducted through field experiments. Lee et al. ( 6 ) conducted a field experiment to evaluate the driving performance of PMVs. For the experiment, a bicycle, a two-wheeled skateboard with a handle, and an electric three-wheeled bicycle were used, and four participants were involved. For the evaluation, safety, steering comfort, behaviors under sudden braking, and pedestrian and bicycle avoidance abilities were assessed. Based on the experimental analyses, it was concluded that the ability of PMVs to overtake and steer is similar to that of a bicycle. The Korea Transport Institute conducted a driving test to investigate and compare the driving performance of PMVs; which included a bicycle, an electric scooter (e-scooter), an electric unicycle (e-unicycle), and a segway. The participants had practiced for two weeks before the test. In the experiment, the ability of PMVs to accelerate, stop, and rotate were assessed. The results showed that the SSD of the PMVs was greater than that of the bicycle at low speeds below 10 km/h, but it was not much different from that of the bicycle at high speeds. The bicycle and e-scooter required a bigger radius to turn as a result of being relatively longer in length ( 27 ). The Centre for Electric Vehicle Experimentation in Quebec conducted an experiment to assess the driving performance of a two-wheeled skateboard with a handle, particularly a segway model. It was intended for those who had over 20 h of driving training before the experiment. The participants were assessed for their maximum acceleration, maximum speed, braking performance, driving performance on grade sections, and turning performance. The results of the experiment showed that the maximum speed of the devices was 20.5 km/h, the maximum acceleration was 7.09 m/s², the SSD was 5.21 m, and the minimum turning radius was 4.57 m ( 28 ). The German Insurance Association ( 11 ) conducted several experiments to investigate the driving characteristics of a segway. The test was conducted after particpants had more than 3 h of driving practice. The experiment was designed to assess SSD, driving over curbs, driving on wet roads, and emergency handling skills. Driving over curbs lower than 1.1 cm was easy, but steering was not comfortable at a height higher than that. The SSD at 20 km/h was between 2.2 and 5.7 m, and it was generally larger than that of the bicycle (2.7 to 4.1 m). There was no issue with driving on wet surfaces, but some participants with short training time lacked the ability to cope with sudden events ( 11 ). The FHWA conducted an experiment on the driving performance of a segway. The experiment was conducted on four segway users and it was concluded that most of the results were in compliance with the design criteria of AASHTO’s bicycle lanes ( 29 ). Jeong et al. ( 30 ) conducted field experiments using a two-wheeled skateboard with a handle and pedals. Tests to assess driving over a curb and on a grade section and on roadways were conducted, and it was found that a device using both manpower and electric power was less burdensome to the battery and motor. Speeds and overtaking behaviors of a segway when facing obstacles were investigated by Miller et al. The amount of lateral encroachment when overtaking stationary obstacles (0.4 m), such as traffic cones, was lower than when overtaking moving obstacles (0.9 m), such as pedestrians ( 31 ). Dowling et al. ( 32 ) conducted experiments at the campus of Macquarie University (Sydney, Australia). The experiments were conducted using e-scooters, segways, and three-wheeled skateboards. A survey was also conducted to investigate user experience and preference before and after using PMVs. Through GPS, the trip routes and average speed were investigated, and videos were also recorded. In the experiments, the maximum speed of the devices was set up to 10 km/h for the safety of those on campus. During the experiments, most users considered it was easy to drive alongside pedestrians on shared paths. Participants had no major issues when driving on slope sections, but they experienced some uncomfortable behaviors while maintaining their balance ( 32 ). Park et al. ( 8 ) reviewed the quantitative safety assessment indicators of the existing modes of transportation and proposed those of the PMVs. The center of the gravity and stability metric were selected as the indicators for the safety of PMVs. In addition, the safety of PMVs was evaluated by using a motion analysis system. Through the safety assessments for three different types of PMVs, each of which had a different number of wheels, it was found that the e-scooter with two wheels exhibited the worst degree of safety ( 8 ).

In all the previous studies, the experiments were conducted with relatively few participants, thus the results were less reliable. Second, the results of experiments conducted with limited types of PMVs would not represent those of all PMVs. Lastly, few studies have investigated the driving performances and overtaking behaviors of the PMVs in environments of mixed traffic representing real road conditions. Therefore, in this study, the driving performances and characteristics of different types of PMVs were examined thoroughly based on field experiments and compared with those of bicycles. Furthermore, other issues on driving behaviors of PMVs, such as overtaking maneuvers, were taken into account as well.

Methodology

In this study, field experiments were conducted to investigate the driving performances and characteristics of various PMVs with funding support from KoROAD. The driving performances of PMVs were analyzed through specified measures and specific behaviors. The feasibility of driving a PMV on bicycle lanes was assessed by device types, and adequate design criteria for PMV lanes were extracted from the corresponding results.

Experiment Design

Objectives of Experiments

To lay the groundwork for permitting PMVs to be driven on bicycle lanes, the driving performances and behavior characteristics were evaluated when PMVs were driven on bicycle lanes, and problems with PMVs were also identified. In addition, the experimental data were analyzed to provide specific grounds for assessing the feasibility of driving PMVs on bicycle lanes by device types. Furthermore, through the analysis results, the design criteria for a new type of road only for PMVs were derived.

Experimental Methods

To fulfill the objective of this study, two experiments were conducted. The first experiment was to assess the driving performance of PMVs on bicycle lanes. The second experiment was composed of four tests to assess the adequacy for PMVs of the design criteria of bicycle lanes and to derive criteria for PMV lanes. The tests in the second experiment included a test for assessing SSD, turning maneuvers, lane width, and driving performance on a grade section. There are many types of PMVs used in South Korea. In this experiment, we tested three types of PMVs: e-scooter, e-unicycle, and segway. The term “segway” originally referred to a personal transporter manufactured by Segway Inc., but is now used widely to indicate a motorized two-wheeled vehicle with a handle. Participants who had been driving PMVs for more than one month were recruited. A total of 13 PMV users participated in the first experiment (five each for the e-scooter and e-unicycle, three for the segway), and 36 participants were engaged in the second experiment (12 for each device). The PMVs owned by each of the participants were used in the experiment. Figure 1 shows the three types of PMV used in the experiments.

Figure 1.

Personal mobility vehicles used in the experiments: (a) electric scooter, (b) electric unicycle, and (c) segway.

All the experiments were recorded as videos using action cameras installed on adjacent utility poles and on a crane raised about 15 m above the ground. The data on PMV driving behavior were extracted and subsequently analyzed. Figure 2 shows the field experiment displays.

Figure 2.

Field experiment display: (a) side view of the field test for assessing the driving performance of the personal mobility vehicles on bicycle lanes in various conditions and (b) top view of the test for measuring the turning maneuvers of the PMVs.

Test of Driving on Bicycle Lanes in Various Conditions

A test was conducted to assess the driving performance of the PMVs on bicycle lanes. It was designed according to the circumstances that the PMV drivers could encounter when driving on bicycle lanes. Four scenarios were developed based on two conditions: mixed traffic conditions and side obstacles, such as curbs and fences, along bicycle lanes. The divided scenarios are shown in Table 2. Scenario 1 is non-mixed with bicycles without fences and curbs in a long tangent section, and scenario 2 is mixed with bicycles without fences and curbs in a long tangent section. Scenario 3 is mixed with bicycles on a long tangent section with curbs, and scenario 4 is mixed with bicycles on a long tangent section with fences. The differences between the results from the first two scenarios were used to assess influence of bicycles on the driving performance of PMVs, and the results of scenario 3 were compared with that of scenario 4 to evaluate the impacts of fences and curbs on PMVs’ driving performance. The average speed values of each PMV were obtained at each section, and the driving characteristics were identified by recorded videos.

Table 2.

Scenario Design of the Driving Track Designed According to the Korean Standard of Bicycle Lanes

Scenario design	Scenario 1	Scenario 2	Scenario 3	Scenario 4
Traffic composition	Personal mobility vehicles (PMVs) only	PMVs and bicycles mixed	PMVs and bicycles mixed	PMVs and bicycles mixed
Fences/curbs	X	X	Curbs	Fences

The track for this experiment is shown Figure 3, and the participants drove counter-clockwise on the track, which consisted of two tangent sections and four curved sections. On the long straight section, fences and curbs were installed in scenarios 3 and 4. Each curved section had a radius within the design criteria of bicycle lanes. There were three curved sections designed using different design speeds.

Figure 3.

Track designed according to the South Korean standard for bicycle lanes.

Test for the SSD

It is important to examine the stopping ability of the vehicles to avoid conflicts and accidents in unexpected situations on the road. SSD is a principal factor in road design criteria, and therefore tests for assessing the SSDs of PMVs were conducted. After the tests, it was evaluated whether the SSDs of the PMVs follow the design standards of bicycle lanes. The testing ground is illustrated in Figure 4. The track is composed of an acceleration section and a stopping section. The participants were required to accelerate on the acceleration section and stop in the stopping section when they received sudden flags. They could modify the speed by checking their own speed on the front display. By using the recorded videos, the SSDs of the PMVs were measured at running speeds of 10, 15, 20, and 25 km/h. However, the SSDs of the segways were assessed only below 20 km/h because of hardware limitations.

Figure 4.

Track layout for the stopping sight distance test.

Test for the Turning Maneuver

Turning sections are relatively difficult to drive on, so drivers could more likely experience difficulties here than in the tangent sections. Furthermore, the turning radius is a major element of road design criteria, similar to the SSD. Therefore, it is necessary to evaluate the driving performance of the PMVs at the turning sections. For this objective, tests for evaluating the turning performance of PMVs were conducted. Then, it was assessed whether the turning radii of the PMVs conform to the design standards of bicycle lanes. The testing ground is shown in Figure 5. The track was divided into an acceleration section and a turning section. The participants were instructed to accelerate on the acceleration section and turn at the turning section on seeing sudden flags. They could modify the speed by checking their own speed on the front display. By using recorded videos, the turning radii for different speeds of each PMV were calculated. The turning radius of each PMV was measured at the running speeds of 10, 15, 20, and 25 km/h. Similar to the SSD tests, in this case also, the turning radii of the segways were not measured over 15 km/h.

Figure 5.

Track layout for the turning radius test.

Test for Driving Performance on a Slope Section

One of the road factors that can cause difficulties while driving is a slope. The driving performance of vehicles is restricted because of the power limitations in the slope section, and this applies to PMVs as well. It is therefore crucial to evaluate the climbing ability of the PMVs on an inclined road section. The test was designed to assess the driving performance of the PMVs on an upward grade section such as can be encountered when driving on roadways. The testing ground is shown in Figure 6. The speeds through the grade section (upward slope: +7%) were measured by analyzing recorded videos.

Figure 6.

Track layout of the test for driving on a slope section.

Test for Adequacy of Bicycle Lanes’ Width for PMVs

PMVs vary in size and have different dimensions when compared with those of bicycles. Especially, horizontal length is a major difference between devices. For example, bicycle and e-scooter have two wheels connected in series, while a segway’s wheels are connected in parallel. The difference in horizontal length is an important issue because roads must provide sufficient width to the vehicles using them. Thus, the driving performance of the PMVs in different lane widths (1.0, 1.5, and 2.0 m) was examined. The test consisted of two courses with different driving speeds (10 km/h, 20 km/h) and turning radii (5 m at 10 km/h, 12 m at 20 km/h). One of the courses is displayed in Figure 7. The participants accelerated their device in the acceleration section and drove through three curved sections. After covering the first course, they moved to the second course and drove through it.

Figure 7.

Track layout for assessing the adequacy of width of bicycle lanes for personal mobility vehicles.

Results

Driving Behavior of PMVs in Bicycle Lane Environments

To investigate the driving performances of the PMVs, experiments with several scenarios were conducted to test whether the driving maneuvers of the PMVs are suitable to the standards of bicycle lanes. The experimentally obtained data were analyzed to examine the driving performances and characteristics of the PMVs.

Running Speeds of the PMVs on Bicycle Lanes

First, the running speeds of the PMVs on bicycle lanes without any roadside obstacle, such as curbs and fences, were analyzed. The speeds of the e-unicycles were relatively high in all the scenarios and in both mixed and non-mixed traffic conditions, whereas the average speed of the segways was the lowest as a result of the mechanical characteristics of these devices. Generally, the speeds of the PMVs were higher in non-mixed conditions than in mixed conditions. This is because the PMVs were interrupted by bicycles moving at lower speeds. These running differences may cause the PMVs to overtake the bicycles in real scenarios.

The average speeds of the PMVs in the curved section with a 12 m radius were higher than in the tangent section. In addition, the average speeds of PMVs in curved sections with radii under 12 m were lower than those in the tangent section except for segways (10 m in the case of segways). This implies that, while driving, the PMV drivers do not feel any difference between the environments of the tangent section and those of the curved section with a 12 m radius (10 m for the segway users). Table 3 shows the results of the tests for running speeds of the PMVs on bicycle lanes.

Table 3.

Average Speed During Driving in Mixed Traffic and Non-Mixed Traffic Environments

Traffic environment	Space mean speeds by sections (km/h)
Traffic environment	Tangent section	Curve (R = 12 m)	Curve (R = 10 m)	Curve (R = 7 m)	Curve (R = 5 m)
E-scooter
Non-mixed	19.59	21.09	19.43	9.45	9.28
Mixed	16.23	18.31	16.14	7.97	7.53
Total	18.80	19.78	17.43	9.04	8.37
E-unicycle
Non-mixed	23.81	26.42	20.32	15.17	14.12
Mixed	22.03	23.32	18.43	12.47	11.10
Total	22.46	24.08	18.89	13.12	11.84
Segway
Non-mixed	18.08	18.20	18.27	13.63	11.18
Mixed	17.34	17.34	18.76	13.64	10.85
Total	16.86	17.30	18.12	13.18	11.13

Overtaking Behavior

The results of overtaking behavior in the long tangent section in scenarios 2, 3, and 4 are shown in Table 4 and Figure 8. In scenario 2, the e-unicycles exhibited the highest frequency of overtaking because of their significantly higher running speeds and more flexible movements, but at the same time, they exhibited a higher number of encroachments onto the next lane. Segways had a similar frequency of overtaking and encroachments, which implies that it was not easy for drivers to overtake within the given width of the bicycle lanes, 1.5 m. PMVs overtook bicycles more than PMVs overtook PMVs. This would be because PMVs have difficulty in overtaking other PMVs in the bicycle lanes as they have relatively greater widths.

Table 4.

Average of Overtaking Behaviors in Long Tangent Section

Classification	Number of overtaking behaviors per trial (Number of encroachments during overtaking per trial)
Classification	E-scooter	E-unicycle	Segway
Scenario 2 (without curbs and fences)
Personal mobility vehicle (PMV) over bicycle	0.068 (0.033)	0.641 (0.062)	0.200 (0.174)
PMV over PMV	0.132 (0.093)	0.484 (0.313)	0.071 (0.071)
Total	0.200 (0.126)	1.125 (0.375)	0.271 (0.245)
Scenario 3 (with curbs)
PMV over bicycle	0.085 (0.078)	0.429 (0.000)	0.333 (0.333)
PMV over PMV	0.078 (0.071)	0.357 (0.000)	0.083 (0.083)
Total	0.163 (0.149)	0.786 (0.000)	0.417 (0.417)
Scenario 4 (with fences)
PMV over bicycle	0.018 (0.018)	0.407 (0.131)	0.000 (0.000)
PMV over PMV	0.000 (0.000)	0.185 (0.262)	0.000 (0.000)
Total	0.018 (0.018)	0.593 (0.393)	0.000 (0.000)

Figure 8.

Overtaking behaviors and encroachments during driving on a long tangent section.

To assess the impact of roadside obstacles (curbs and fences) on the overtaking behavior of the PMVs, the results obtained in scenario 2 were compared with those of scenarios 3 and 4. The number of overtaking events for all the PMVs in scenario 2 was significantly greater than those in scenario 3 (driving bicycle lanes with curbs) and scenario 4 (driving bicycle lanes with fences), except for the segways. This might be because PMVs have difficulty in overtaking other modes because of insufficient space to overtake in lanes with curbs and fences. Analogous to the results of overtaking behaviors in scenario 2, the e-unicycles exhibited the highest frequency of overtaking compared with the others in scenario 3 and 4. Furthermore, they frequently encroached onto the next lane during overtaking. Segways had a relatively low frequency of overtaking, and most of them could not overtake other modes without any encroachment onto the next lane because of the widths of the devices.

Driving Performance of PMVs for Road Design Criteria

The SSDs and turning radii of the PMVs were examined to establish design standards for PMV lanes, and these design standards were compared with those of the bicycle lanes. The comparison elucidated whether the driving performance of PMVs was suitable for the design standard of bicycle lanes. If there were no difference between the driving performance of PMVs and bicycles, or if the driving performance of the PMVs conformed to the design criteria of bicycle lanes, then it would be reasonable to conclude that driving PMVs on bicycle lanes is appropriate. In addition, the driving performance of PMVs in an upward grade section was appraised to investigate any difficulty when driving in such a section. Moreover, experiments were conducted to identify the lane widths suitable for PMVs.

SSD Test

This test was conducted to evaluate if the SSDs of PMVs conform to the design standards of bicycle lanes. Furthermore, the standards for PMVs lanes were derived from the test. For these purposes, SSDs at four different speeds, 10, 15, 20, and 25 km/h, were measured through this test. As shown in Table 5, at 10 and 15 km/h, the length of SSDs was in the order of: segways, e-unicycles, and e-scooters. Above 20 km/h, the SSDs of e-unicycles were longer than those of e-scooters.

Table 5.

Comparison of Stopping Sight Distance of Personal Mobility Vehicles with Design Standards of Bicycle Lanes

Category				E-scooter	E-unicycle	Segway	Design standard of bicycle lanes
10 km/h	Normality test statistic (p-value)			0.72 (0.000***)	0.95 (0.010*)	0.96 (0.050)	9
	Statistical inference test^a	t-test^b	Mean	2.57	2.72	3.44
		t-test^b	Statistic (p-value)	NA	NA	−36.4 (0.000***)
		Wilcoxon signed-rank test^c	Median	2.10	2.71	3.48
		Wilcoxon signed-rank test^c	Statistic (p-value)	0 (0.000***)	0 (0.000***)	NA
15 km/h	Normality test statistic (p-value)			0.75 (0.000***)	0.99 (0.900)	0.99 (0.800)	14
	Statistical inference test^a	t-test^b	Mean	4.45	4.94	5.12
		t-test^b	Statistic (p-value)	NA	−89.2 (0.000***)	−66.7 (0.000***)
		Wilcoxon signed-rank test^c	Median	3.80	4.88	5.07
		Wilcoxon signed-rank test^c	Statistic (p-value)	0 (0.000***)	NA	NA
20 km/h	Normality test statistic (p-value)			0.79 (0.000***)	0.89 (0.000***)	NA	20
	Statistical inference test^a	t-test^b	Mean	6.20	8.09	NA
		t-test^b	Statistic (p-value)	NA	NA	NA
		Wilcoxon signed-rank test^c	Median	5.25	7.90	NA
		Wilcoxon signed-rank test^c	Statistic (p-value)	0 (0.000***)	0 (0.000***)	NA
25 km/h	Normality test statistic (p-value)			0.90 (0.000***)	0.96 (0.030*)	NA	27
	Statistical inference test^a	t-test^b	Mean	9.02	11.57	NA
		t-test^b	Statistic (p-value)	NA	NA	NA
		Wilcoxon signed-rank test^c	Median	8.35	11.67	NA
		Wilcoxon signed-rank test^c	Statistic (p-value)	0 (0.000***)	0 (0.000***)	NA

Inference test for difference between experiment results of each driver and design standard of bicycle lanes.

If the data group was normally distributed, a t-test was conducted.

If the data group was not normally distributed, the Wilcoxon signed-rank test was conducted.Note: NA = Not available.

p-value < 0.05.

***

p-value < 0.001.

The SSDs of the PMVs were statistically compared with the design standards of bicycle lanes. A t-test and the Wilcoxon signed-rank test were used to compare the SSDs of PMVs with the bicycle lanes’ design standards. The t-test was used for data groups normally distributed, whereas the Wilcoxon signed-rank test was used for data groups that were not normally distributed. The Shapiro-Wilk test was used to test the normality of each data group. The results of Shapiro-Wilk tests showed that three data groups—SSDs of e-unicycles for 15 km/h and for segways for 10 km/h and 15 km/h—were normally distributed. Therefore, the t-test was used for comparing the average value of the three groups’ SSD with the design standards of bicycle lanes, and the Wilcoxon signed-rank test was used for other groups. The comparison results showed that the average SSDs of the PMVs were statistically smaller than the bicycle lanes’ standards for all speeds. In addition, the results indicated that the PMVs did not exhibit any trouble during stopping in the bicycle lanes.

Turning Radius Test

This test was conducted to examine whether the PMVs’ turning radius conforms to the design standards of bicycle lanes. For this purpose, the turning radii at four different speeds: 10, 15, 20, and 25 km/h, were measured. Under 25 km/h, the e-unicycle had a significantly larger turning radius compared with the other PMVs, as shown in Table 6. However, at 25 km/h, the average turning radius for the e-scooters was slightly larger than that of the e-unicycles.

Table 6.

Comparison Turning Radius of Personal Mobility Vehicles with Design Standards of Bicycle Lanes

Category				E-scooter	E-unicycle	Segway	Design standard of bicycle lanes
10 km/h	Normality test statistic (p-value)			1.0 (0.004**)	0.9 (0.000***)	0.8 (0.000***)	5
	Statistical inference test^a	t-test^b	Mean	5.10	6.30	6.06
		t-test^b	Statistic (p-value)	NA	NA	NA
		Wilcoxon signed-rank test^c	Median	4.91	5.66	5.56
		Wilcoxon signed-rank test^c	Statistic (p-value)	882 (0.492)	1,169 (0.992)	1,298 (0.998)
15 km/h	Normality test statistic (p-value)			0.9 (0.002**)	0.9 (0.000***)	1.0 (0.146)	7
	Statistical inference test^a	t-test^b	Mean	6.27	7.84	6.23
		t-test^b	Statistic (p-value)	NA	NA	−3.6 (0.000***)
		Wilcoxon signed-rank test^c	Median	5.77	7.03	6.37
		Wilcoxon signed-rank test^c	Statistic (p-value)	424 (0.000***)	1,070 (0.875)	NA
20 km/h	Normality test statistic (p-value)			0.9 (0.006**)	0.8 (0.000***)	NA	12
	Statistical inference test^a	t-test^b	Mean	8.05	8.68	NA
		t-test^b	Statistic (p-value)	NA	NA	NA
		Wilcoxon signed-rank test^c	Median	7.70	7.68	NA
		Wilcoxon signed-rank test^c	Statistic (p-value)	8 (0.000***)	283 (0.000***)	NA
25 km/h	Normality test statistic (p-value)			1.0 (0.166)	0.9 (0.004**)	NA	18
	Statistical inference test^a	t-test^b	Mean	9.62	8.39	NA
		t-test^b	Statistic (p-value)	−32.0 (0.000***)	NA	NA
		Wilcoxon signed-rank test^c	Median	9.29	8.35	NA
		Wilcoxon signed-rank test^c	Statistic (p-value)	NA	0 (0.000***)	NA

Inference test for difference between experiment results of each driver and design standard of bicycle lanes

If the data group was normally distributed, a t-test was conducted.

If the data group was not normally distributed, the Wilcoxon signed-rank test was conducted.Note: NA = Not available.

p-value < 0.01.

***

p-value < 0.001.

The t-test and Wilcoxon signed-rank test were also used to compare the turning radii of PMVs at each speed with the design standards of bicycle lanes. The Shapiro-Wilk test was used to test the normality of each data group before the comparison. The results of Shapiro-Wilk tests showed that two data groups—e-scooters’ turning radius for 25 km/h and segways’ turning radius for 15 km/h—were normally distributed. Therefore, the t-test was used for comparing the average value of the two groups’ turning radii with the design standards of bicycle lanes, and the Wilcoxon signed-rank test was used for other groups. Over 15 km/h, the average turning radii of the e-scooters and e-unicycles were statistically smaller than the standards. However, at 15 km/h, the turning radius of the e-unicycle was larger than the standards, and the turning radii of all the PMVs were above the standard value at 10 km/h. The design speed is the speed at which the driver can maintain comfort in any section of the road. In addition, the speed limit of the road should be set equal to or less than the design speed. Therefore, if the vehicle’s performance does not meet the design criteria according to the design speed when driving at a speed equivalent to the design speed, it would mean there is a possibility of problems when the vehicle drives on the road designed to the design standards. The experimental results indicated that if the PMV is driven at a speed equivalent to the design speed when turning at a low design speed (10 km/h, 15 km/h), it can turn at a radius of rotation larger than the design standard, which meant that the PMV might have a specific problem when turning on a bicycle lane. Therefore, it may be inappropriate for PMVs to drive on bicycle lanes, especially at the turning sections, when driving at low speeds.

Driving on Grade Section Test

This test was conducted to the ability of the PMVs when driving on an upward slope section. The participants were required to drive their PMVs on an uphill road with a slope of 7% and a length of 340 m. During the test, their driving speeds were measured every 10 m to investigate whether the upward slope adversely affected their speeds. To evaluate the problems that may occur when PMVs are driven on a slope section, it is necessary to focus on the results of the devices with poor performance rather than those with good performance. Therefore, the 15th percentile of the speeds at each point was selected as the representative value to embrace the lower ranks of the results for each PMV, and the outputs are as shown in Figure 9. PMVs accelerated to about 50 m and maintained each constant level of speed. Among the PMVs, the 15th percentile speed of the e-unicycles was higher than that of the other PMVs at every point, and the segways’ speed was the lowest. According to the bicycle lanes’ standards for uphill sections, the length of an uphill section with a slope 7% is limited to 120 m. As there was no impact of the slope on the speeds of PMVs over 120 m, PMVs would not have any difficulty on the slope section less than 7%.

Figure 9.

Driving speeds of personal mobility vehicles in a slope section by distance (15th percentile).

Lane Width Test

The lane width test was conducted to evaluate whether the width of the bicycle lanes is suitable for PMVs. The test lane consisted of two courses with different driving speeds and rotational radii, and there were three turning sections with different lane widths in each course. The participants were required to drive their PMVs through each course. During the test, the degree to which the PMVs deviated from the center of the road was measured at the start, middle, and end points of each turning section. The lateral placement (LPM) was set to be negative when the PMVs moved away from the center to the left, and positive when they moved away to the right. Furthermore, the LPM was set to 0 when the PMVs did not deviate from the center of the road, −1 when the left line of the lane and the left wheel met, and +1 when the right line of the lane and the right wheel touched. The results of the lane width test are shown in Table 7.

Table 7.

Lateral Placement (LPM) of Personal Mobility Vehicles on the Lane Width Test Tracks

Speed	Curved section number	Lane width	Direction	Radius	Average LPM of PMVs on curved section (Average LPM of PMVs in the centrifugal direction)
Speed	Curved section number	Lane width	Direction	Radius	E-scooter	E-unicycle	Segway
10 km/h	No. 1	1.0m	Counter-clockwise	5 m	0.04 (0.33)	0.22 (0.52)	0.03 (0.42)
	No. 2	1.5m	Clockwise	5 m	−0.11 (0.33)	−0.09 (−0.30)	0.03 (−0.25)
	No. 3	2.0m	Counter-clockwise	5 m	0.04 (0.25)	0.08 (0.29)	0.06 (0.21)
20 km/h	No. 1	1.0m	Counter-clockwise	12 m	−0.06 (0.52)	0.10 (0.47)	0.05 (0.31)
	No. 2	1.5m	Clockwise	12 m	−0.19 (−0.34)	0.11 (−0.23)	−0.01 (−0.27)
	No. 3	2.0m	Counter-clockwise	12 m	0.07 (0.27)	0.08 (0.22)	0.07 (0.23)

The LPM values of the PMVs were expected to be in the direction of the centrifugal force. However, according to the average values of the PMVs’ LPM for each curved section, not every value matched with the centrifugal direction, and did not exceed −1 or +1, which suggests that the bicycle lane width standards are sufficient for the PMVs to drive in. The e-scooters were not in the centrifugal direction in the No. 1 curved section for 20 km/h, and the e-unicycles and segways showed the same trend in the No. 2 curved section for 20 km/h and10 km/h, respectively. These results explain that PMVs are rarely affected by the width of the bicycle lanes while driving through the turning sections of these lanes. As a result of measuring the values of PMVs’ LPM in the centrifugal direction only, it was found that PMVs were stable enough not to have excessive values of LPM, at most an absolute value of 0.52. Through assessing the LPM of PMVs in curved sections designed according to the bicycle lane standards, it was found that PMVs are adequate to drive on the road with widths of bicycle lane standards.

Summary of Results

The field experiment results of PMVs with regard to maneuverability on bicycle lanes are as follows:

Driving performance of PMVs in the bicycle lanes environment: PMVs had difficulties in overtaking in a given width in the test. The e-unicycle overtook other modes more than other PMVs, but at the same time, it encroached on the next lane much more frequently. The segways exhibited a similar frequency of overtaking and encroachment because of their large widths that enabled overtaking within the bicycle lanes’ width. These results show that it is not easy for PMVs to overtake in bicycle lanes.

SSD: All types of PMVs for overall speeds had lower SSD values than the values of bicycle lane SSD standards. This indicates that PMVs could stop within the distance of the bicycle lane standards.

Turning radius: Over 15 km/h, the turning radii of the e-scooters and e-unicycle were smaller than those of the bicycle lane standards. However, the values were higher than the standards under 20 km/h, which implies that the PMVs would have trouble when turning on a curved section of the bicycle lanes designed with design speeds under 20 km/h.

Driving performance of the PMVs on a slope section: Among the 15th percentile of driving speeds of the PMVs, the e-unicycle was faster than the others at all the points, followed by the e-scooter and then the segway. The PMVs maintained their speeds on a 7% slope section, indicating that the PMVs are not much affected by the slope on the roads.

LPM of the PMVs in a curved section: For the lane width for the 1.0, 1.5, and 2.0 m, the participants maintained their devices’ lateral position within each lane width. This shows that PMVs can be driven within the bicycle lanes’ width.

According to the results of analyses, it was not difficult for PMVs to drive on bicycle lanes. However, PMV users would have some trouble when turning at curved sections, especially with low design speeds. Moreover, PMVs are too wide to overtake other modes within the width of bicycle lanes.

Design Criteria of PMV Lanes

Through the previous experimental results, it was confirmed that it was not difficult to drive PMVs on bicycle lanes. However, several problems were found, such as the turning radius at low speeds and overtaking behavior. Therefore, it was judged that it was difficult for PMVs to drive completely safely on the bicycle lanes. It was considered to allocate road space for PMVs only—PMV lanes—to solve the problem. To create a road only for PMVs, it is necessary to set design criteria suitable for the driving performance of PMVs. To construct design criteria for PMV lanes, it is necessary to scrutinize the design criteria of bicycle lanes because bicycles are most similar to PMVs among all transport modes. In the bicycle lanes’ design criteria, the SSD and turning radius by design speeds are presented, and the length limits for slope sections with different slope level are also proposed. Additionally, the bicycle lane widths are specified according to the types of the bicycle lanes. Accordingly, this study was intended to present the design standards for the following criteria for PMV lanes: SSD and turning radius for each design speed, width of PMV lanes, and installation of overtaking lanes in the PMV lanes.

SSD and Turning Radius

In the preceding analysis, the SSD and turning radius of the PMVs were measured, and the average values of each were compared with the design standards of bicycle lanes. As design standards, it is dangerous to use the average values of SSD and turning radius because PMV lanes criteria cannot embrace devices with low driving performances. Therefore, in this study, the design standards of the SSD and turning radius were set using the 85th percentile value. That is, the minimum integer value equal to or greater than the 85th percentile value of the SSD and turning radius was selected for the design standards. The corresponding results are shown in Table 8.

Table 8.

Design Standards of Stopping Sight Distance (SSD) and Turning Radius for Personal Mobility Vehicle (PMV) Lanes

		SSD (m)			Turning radius (m)
Category		E-scooter	E-unicycle	Segway	E-scooter	E-unicycle	Segway
85th percentile values for each test	10 km/h	3.20	3.33	4.80	6.24	9.42	7.80
	15 km/h	6.32	5.82	6.20	7.75	11.36	7.95
	20 km/h	7.95	9.14	-	9.54	11.42	-
	25 km/h	12.28	13.79	-	11.67	12.04	-
Design standards for PMV lanes	10 km/h	5			10
	15 km/h	7			12
	20 km/h	10			12
	25 km/h	14			13

Lane Width and Length Limit of Slope Section

The analysis of the driving performances of the PMVs in the uphill slope section confirmed that a constant running speed was maintained on the 7% uphill slope. Unlike bicycles that use human power, PMVs do not decrease their vehicle speed on the slope as they run on electricity. Therefore, unlike bicycle lanes that suggest a slope limit length, PMV lanes do not need a length limit on the slope sections.

The lane width test results confirmed that PMVs can be driven within the width of the bicycle lanes when traveling in the curved sections of these lanes. Therefore, a width standard needs to be set for the PMV lanes, according to the standards of the bicycle lanes, because of the similarity between bicycles and PMVs in appearance and driving performance.

Passing Lanes

Because the PMVs vary in speed by types, if various types of PMVs and other means are mixed in traffic, then overtaking occurs because of speed differences. However, delays may occur if overtaking is not possible. The analysis of the overtaking behavior of the PMVs revealed that PMVs, especially those with large widths, often encroach the road line when overtaking. This implies that a sufficient width for overtaking is not secured within a given width of the lane. Therefore, passing lanes for overtaking need to be installed at regular intervals to prevent delays caused by the impossibility of overtaking.

Conclusion

PMVs are expected to be the future mode of travel for “first-mile and last-mile trips” in urban areas. Accordingly, the purposes of using PMVs are diversifying in South Korea. However, PMVs were initially not defined in the Korean regulations, and the corresponding laws were only recently revised to include PMVs. As a result, information about adequate road environments for safe and convenient driving was not provided, and most PMV users drove on pedestrian paths instead of the roadways to avoid collision with cars and other heavy vehicles. As a result, numerous accidents involving PMVs were reported in South Korea. To mitigate these issues, the government allowed PMVs to be driven on bicycle lanes, after identifying the problems of driving PMVs on these lanes. Additionally, creating separate lanes only for PMVs was considered. The aforementioned improvement plan adopted by the government was formulated based on the findings of our study, in which we assessed whether PMVs could be driven on bicycle lanes through field experiments and analysis of the various driving behaviors of PMVs under different real-life conditions. The experimental study data were also used to construct the design criteria for PMV lanes.

The field experiments were designed specifically for creating a driving environment for PMVs. The results obtained from the tests performed to evaluate the driving performance of PMVs in the bicycle lane environments confirmed that it is difficult for PMVs to overtake within the width of bicycle lanes. Further, the actual SSDs and turning radii of PMVs were identified and found to be generally smaller than those of the standard bicycle lane design. While evaluating the driving performance in the slope section, it was observed that no speed reduction occurred in the 7% slope section, indicating that, unlike bicycles, PMVs do not need to limit the distance in the slope section. In addition, the LPM analysis of the PMVs in the curved section revealed that PMVs can travel safely within a width of 1 to 2 m.

Based on the results of the experiments conducted in the present study, it can be concluded that PMVs are largely safe in real-life driving and exhibit only negligible issues when driving on bicycle lanes. Several problems were observed when using PMVs in bicycle lanes. Consequently, creating a new type of lane only for PMVs was considered, and the criteria for the lanes were derived from the results of the field experiments. First, the SSD and turning radius standards were decided using the 85th percentile values. Then, the adequate lane width for the PMV lanes was evaluated. The lane width test results showed that the standard lane width of bicycle lanes (1.5 m for one direction) is sufficient for PMV lanes as well. Furthermore, it was found that the length of the slope section need not be limited because the ability of the PMVs to maintain certain speeds was not affected by the slope. The overtaking behaviors were also analyzed, and the corresponding results indicated that PMVs exhibit difficulties in overtaking or being overtaken within the width of the bicycle lanes. Therefore, it is necessary to build sections for overtaking in bicycle lanes. In addition, the lack of any protection in unexpected situations is a substantial drawback of the PMVs, and accordingly, the users of this transport mode need to wear protective equipment to safeguard themselves when facing unexpected events like collisions.

After the Road Traffic Act was revised, PMV users could choose to drive on the road (in car lanes) or in bicycle lanes. In addition, the reformed Road Traffic Act included only e-scooters and segways as PMVs. It was appropriate to allow PMVs to be driven on bicycle lanes based on the results of this study. However, as PMV users were allowed to drive on bicycle lanes, bicycle users were more likely to experience inconvenience than before the change of the law. Therefore, it is necessary to find a way to solve any problem for bicycle users when using bicycle lanes with PMVs.

For further research, experiments should be conducted with a greater number of participants and more types of PMVs under various experimental conditions such as quality of road surface. Additionally, user experiences should be investigated to reflect PMV users’ needs in the future. Moreover, it is necessary to investigate any problems, such as accidents or collision possibilities, between PMVs and bicycles in bicycle lanes. Because we do not know the causes of all PMV accidents, more accident analyses should be conducted to make safer driving environments for PMVs. Furthermore, the solutions for dealing with the problems should be made through related research.

Footnotes

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: S. Kim, M. Myeong, Y. Jang, D. Lee; data collection: S. Kim, S. Hwang; analysis and interpretation of results: S. Kim, S. Hwang, M. Myeong, D. Lee; draft manuscript preparation: S. Kim, D. Lee. 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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is financially supported by Korea Ministry of Land, Infrastructure and Transport (MOLIT) as “Innovative Talent Education Program for Smart City.”

ORCID iDs

Sunhoon Kim

Sooncheon Hwang

References

Kim

I. H.

Choi

K. M.

Jeon

J. I.

A Survey on Riding Characteristics and Helmet Wearing Conditions of Bicycle and PMV (Personal Mobility Vehicle) Riders. Fashion & Textile Research Journal, Vol. 20, 2018, pp. 63–74.

Han

D. J.

Kim

E. C.

M. K.

Analysis of Severity Factors in Personal Mobility (PM) Traffic Accidents. Journal of Korean Society of Transportation, Vol. 38, 2020, pp. 232–247.

Myeong

M. H.

Song

S. Y.

Choi

M. S.

A Study on the Safety Measures for the Use of New Transportation. Korea Road Traffic Authority Traffic Science Institute, Wonju, 2016.

Kim

J. Y.

A Study on Consumer Protection of Personal Mobility. Korean Consumer Agency, Eumseong, 2017.

Park

J. J.

A Study on the Legislation Related to Personal Mobility. Law & Policy Review, Vol. 23, 2017, pp. 77–106.

Lee

S. I.

Kim

S. H.

Kim

T. H.

A Comparison Study on the Risk and Accident Characteristics of Personal Mobility. Journal of the Korean Society of Safety, Vol. 32, 2017, pp. 151–159.

Woo

J. W.

A Study on Standard Test Method for Evaluation of Driving Performance for Personal Mobility. Proceedings of the Korean Society of Automotive Engineers Annual Conference, Vol. 2016, No. 5, 2016, pp. 1002–1004.

Park

B. J.

Roh

C. G.

Kim

J. S.

A Study on Driving Safety Evaluation Criteria of Personal Mobility. The Journal of the Korea Institute of Intelligent Transportation Systems, Vol. 17, 2018, pp. 1–13.

Statistical Analysis of Traffic Accidents in 2021. Korea Road Traffic Authority, 2022. https://taas.koroad.or.kr/web/bdm/srs/selectStaticalReportsList.do?menuId=WEB_KMP_IDA_SRS_TAA. Accessed November 11, 2022.

10.

Statistical Analysis of Traffic Accidents in 2020. Korea Road Traffic Authority, 2021. https://taas.koroad.or.kr/web/bdm/srs/selectStaticalReportsList.do?menuId=WEB_KMP_IDA_SRS_TAA. Accessed November 11, 2022.

11.

Statistical Analysis of Traffic Accidents in 2019. Korea Road Traffic Authority, 2020. https://taas.koroad.or.kr/web/bdm/srs/selectStaticalReportsList.do?menuId=WEB_KMP_IDA_SRS_TAA. Accessed November 11, 2022.

12.

Statistical Analysis of Traffic Accidents in 2018. Korea Road Traffic Authority, 2019. https://taas.koroad.or.kr/web/bdm/srs/selectStaticalReportsList.do?menuId=WEB_KMP_IDA_SRS_TAA. Accessed November 11, 2022.

13.

Statistical Analysis of Traffic Accidents in 2017. Korea Road Traffic Authority, 2018. https://taas.koroad.or.kr/web/bdm/srs/selectStaticalReportsList.do?menuId=WEB_KMP_IDA_SRS_TAA. Accessed November 11, 2022.

14.

Lee

J. W.

Kim

G. H.

Jung

S. S.

Park

J. S.

Sim

H. S.

Kim

W. T.

A Study on the Crash Simulation Between Personal Mobility Riders and Other Road Users. Proc., Altair Technology Conference 2017, Seoul, 2017.

15.

Kim

G. H.

The Analysis of Injury Risk for the Type of Accident by Personal Mobility. Journal of Auto-Vehicle Safety Association, Vol. 12, 2020, pp. 6–14.

16.

Kühn

Grabolle

Assessing the Safety Characteristics of the Segway. German Insurance Association, Berlin, 2009.

17.

Han

S. Y.

Lee

C. G.

Yun

I. S.

Yoon

Y. I.

J. P.

Analysis of PM (Personal Mobility) Traffic Accident Characteristics and Cause of Death. The Journal of the Korea Institute of Intelligent Transportation Systems, Vol. 20, 2021, pp. 100–118.

18.

Hong

U. P.

Measures to Improve Legislation Regarding Personal Mobility. Legislative Issue Brief Issue, Vol. 14, 2016, pp. 48–55.

19.

Shin

H. C.

Jung

K. O.

Park

S. Y.

A Study on Improving Legislation for Promoting Utilization of Personal Mobility and Its Safety Management. The Korea Transport Institute, Sejong, 2020.

20.

Manual for the Installation and Maintenance Manual of Bicycle Facilities in Seoul. Seoul Metropolitan Government, South Korea, 2016.

21.

Guidelines for the Installation and Management of Bicycle Facilities 2019. Ministry of Land, Infrastructure and Transport, Sejong, 2019.

22.

Guidelines for the Installation and Management of Bicycle Facilities 2020. Ministry of Land, Infrastructure and Transport, Sejong, 2020.

23.

Yoo

K. S.

Hong

J. Y.

Kim

O. S.

Park

S. H.

Strategies to Cope with the Spread of Personal Mobility. The Seoul Institute, South Korea, 2017.

24.

Federal Highway Administration. Characteristics of Emerging Road and Trail Users and Their Safety. Publication FHWA-HRT-04-104. FHWA, U.S. Department of Transportation, Washington, D.C., 2004.

25.

Yang

Jin

Hou

Potential Risk and Its Influencing Factors for Separated Bicycle Paths. Accident Analysis and Prevention, Vol. 87, 2016, pp. 59–67.

26.

Ando

Measuring the Acceptability of Self-Balancing Two-Wheeled Personal Mobility Vehicles. Journal of the Eastern Asia Society for Transportation Studies, Vol. 10, 2013, pp. 444–453.

27.

Shin

H. C.

Lee

J. Y.

Kim

S. L.

Study on the Improvement of Laws and Regulations for Personal Mobilities. The Korea Transport Institute, Sejong, 2016.

28.

Lavallée

Pilot Project for Evaluating Motorized Personal Transporation Devices. Centre for Electric Vehicle Experimentation in Quebec (CEVEQ), Quebec, Canada, 2004.

29.

Landis

B. W.

Petritsch

T. A.

Huang

H. F.

Characteristics of Emerging Road Users and Their Safety. Publication FHWA-HRT-04-103. Federal Highway Administration, McLean, VA, 2004.

30.

Jeong

S. H.

Takahara

Takahashi

Hiroi

Matsumoto

Driving Performance and User’s Evaluation of Self-Balancing Personal Mobility Vehicle with a Pedal Driving Mechanism. International Journal of Advanced Robotic Systems, Vol. 12, 2015, p. 87. https://doi.org/10.5772/60719.

31.

Miller

Molino

J. A.

Kennedy

J. F.

Emo

A. K.

Segway Rider Behavior: Speed and Clearance Distance in Passing Sidewalk Objects. Transportation Research Record: Journal of the Transportation Research Board, 2008. 2073: 125–132.

32.

Dowling

Irwin

J. D.

Faulks

I. J.

Howitt

Use of Personal Mobility Devices for First-and-Last Mile Travel: The Macquarie Ryde Trial. Proc., 2015 Australasian Road Safety Conference, Gold Coast, Australia, 2015.