Use of Mobile Sensing Data for Assessing Vibration Impact of E-Scooters with Different Wheel Sizes

Abstract

With the growing demand of shared electric scooters (e-scooters) for short-distance trips in urban areas, their safety issues have received significant attention from various stakeholders. In general, e-scooter riders encounter more vibrations compared with users of other transportation modes such as bicycles that typically have larger wheels and better suspension systems. Such riding experience may lead to discomfort, affect riders’ health, and increase riding risk. Intuitively, increasing e-scooters’ wheel sizes may provide safer and more comfortable riding experience. However, this assumption has not been well tested. The main objective of this paper is to curtail this gap by quantitatively assessing the impacts riders experienced, through the use mobile sensing data. Specifically, a mobile sensing platform was deployed on e-scooters with different wheel sizes to continuously measure encountered vibrations. Then, the instrumented e-scooters were ridden on routes with different pavement materials. Finally, the collected data were further processed and analyzed to evaluate various riding issues in each riding scenario. The comparative results suggest that e-scooters with larger wheels can efficiently alleviate vibrations during a ride compared with those with smaller wheels. To improve riding experience and safety, it is beneficial to use larger-wheel e-scooters, given other similar riding conditions.

Keywords

electrical scooters mobile sensing sidewalk micro-mobility sensor data analytics roadway infrastructure safety

The rapid evolutions of shared mobility systems have drastically changed the urban mobility landscape. Shared electric scooters (e-scooters), as one of the emerging forms of micro-mobility, quickly flooded the streets in many cities. For example, Washington, D.C. was served by four e-scooter vendors (i.e., Lime, Veo, Spin, and Lyft), for a total of 8,220 e-scooters being allowed to zip around the city from January 1, 2023, through December 31, 2024 ( 1 ) The dockless and portable characteristics have made e-scooters a competitive option for short-distance trips (e.g., trip distance ≤ 2 miles), which accounted for about one thrid of vehicle trips based on 2017 National Household Travel Survey ( 2 ). E-scooters can be ridden on sidewalks and streets, which allows riders to flexibly customize their routes. In addition, e-scooters can be picked up or dropped off at various sites (including designated parking spots and open spaces), which minimizes travelers’ walking distances. Amid the pandemic, many travelers used e-scooters as an individualized mode to commute, allowing them to maintain considerable social distances from each other. Such growing demand has brought new challenges to e-scooter vendors of continuously providing greener, healthier, and safer riding experiences.

Despite the convenience, one critical challenge for e-scooter riders is that they always need to navigate a complex riding environment. Unlike vehicles that are often constrained in disciplined lanes, e-scooters need to frequently change routes planned on shared facilities such sidewalks and bike lanes. According to Ma et al. ( 3 ), e-scooter policies and riding guidelines are still evolving in many cities. In many places, e-scooters can be ridden on public roads and mixing with other transport modes can be disruptive ( 4 ). On one hand, e-scooter riders are vulnerable when riding in vehicle lanes; on the other hand, they may hit pedestrians while riding on sidewalks. Meanwhile, because of the lack of crash data, the risk and conflicts of riding e-scooters can be underestimated. According to Yang et al. ( 5 ), fatal e-scooter-involved crashes have been reported in several places. Many of the reported fallings were found on sidewalks with narrow riding space and uneven pavements ( 6 , 7 ). Sidewalks with poor pavement conditions are not recommended for e-scooter riders because the wheels can be easily caught by potholes or cracks. Nevertheless, many cities lack up-to-date information on the quality of the sidewalks for planning the most rational routes for riders.

Besides e-scooter riders and riding environments, the e-scooter devices’ physical structures also play important roles in e-scooter safety. For example, Ma et al. ( 8 ) developed a mobile sensing platform for quantifying vibrations and the approaching obstacles during an e-scooter ride. It is suggested that the abrupt acceleration and frequent vibrations of e-scooters may lead to uncomfortable and risky riding compared with bikes, which traditionally have larger wheels. This raises the question of whether the larger the e-scooter wheels, the more comfortable the riders will feel. Such concerns have also been noticed by e-scooter vendors. For example, according to Toll ( 9 ), Lime introduced new e-scooter devices with 10 in. wheels to provide smoother riding behaviors than previous devices with 8 in. wheels. The impact of wheels on riding experience has not yet been well assessed and the performances might vary across different road facilities. To fill this gap, this study extended the vibration detection methods from previous work by collecting data using mobile sensing technologies. Then, both small-wheel and large-wheel e-scooters were evaluated in multiple sidewalk scenarios with different facility conditions.

Literature Review

Since the rapid distribution of shared e-scooters, it has been widely discussed where and how e-scooters should be ridden. For example, e-scooters are frequently used for short-distance trips. E-scooters can be an option for connecting to metro stations ( 10 ). With their portable and dockless characteristics, e-scooters can be ridden in various places, which brings new challenges in operations and safety concerns ( 11 , 12 ). Gössling ( 13 ) investigated the challenges associated with the introduction of e-scooters in 10 major cities, based on a content analysis of local media reports. Tuncer et al. ( 14 ) analyzed 171 e-scooter riders from camera records and identified risky riding behaviors such as riding opposite traffic flow and running red lights. As less than 15% of riders wear helmets, they can easily get severe injuries in crashes occurring on streets and sidewalks. A survey of 591 responses by Fitt and Curl ( 15 ) sheds lights on how e-scooters can be used. It was found that, compared with main roads, more e-scooter riders prefer quiet sidewalks, to avoid crashing with vehicles. Nevertheless, there are also safety concerns about riding e-scooters on sidewalks. As reported by Sikka et al. ( 16 ), e-scooters may hit pedestrians on sidewalks. Impacted by such incidents on sidewalks, some cities may eliminate or prohibit e-scooters from sidewalks. For example, Ma et al. ( 3 ) examined municipal guidelines in 80 U.S. cities with e-scooter programs, about half of which restricted e-scooter riders from riding on sidewalks. There is a demand for more efforts to explore how to provide safer riding environments for e-scooter riders.

Besides the conflict with other transportation modes, pavement conditions on sidewalks are also closely related to e-scooter safety. Cano-Moreno et al. ( 17 ) identified that e-scooter vibrations primarily affect the human health and comfort of e-scooter riders. Strong vibrations are expected when riding e-scooters on sidewalks with poor pavements. For example, Austin Public Health ( 18 ) interviewed 125 e-scooter riders about their safety concerns with their riding experiences. Among the responses, 50% believed that surface conditions such as potholes or cracks contributed to their injuries. However, the surveying methods can be considered subjective and biased because of the small sample sizes.

Quantitative analysis sheds light on actual performances of e-scooters. Zou et al. ( 19 ) first provided descriptive statistics on e-scooter share trips, followed by an exploratory analysis of trip trajectories conjoined with street link level features. It was found that arterials and local streets with heavy traffic are the most popular facilities used by e-scooters. Besides trip-level analysis, researchers also collect vibration data to provide stronger reasoning on the impacts of pavement conditions. For example, Ma et al. ( 8 ) developed a portable device with motion tracking sensors installed. Vibration data were collected while riding e-scooters equipped with this device. The data were paired with GPS measurements to identify hotspots where riders encountered strong and frequent vibrations.

On the other hand, besides the pavement conditions, the physical characteristics of e-scooter devices may also influence riding experience. Compared with conventional bicycles with large wheels, e-scooter riding is often described as ‘twitchy,’‘wobbly,’ and ‘less stable’ ( 20 ). Wheel size also plays an important role toward safer e-scooter design. Iftakhar ( 21 ) conducted pothole tests between e-scooters with 8 in. and 16 in. wheels. It was found that potholes had less impact on the e-scooters with larger wheels. Nowadays, many e-scooter vendors (e.g., Lime and Bird) have upgraded existing e-scooters from small wheels toward larger wheels. However, no previous research has extensively tested the impact of wheel size when riding e-scooters on sidewalks. To fill this gap, this study takes advantage of mobile sensing technologies to compare the performances between small-wheel and large-wheel e-scooters, to provide a more objective evaluation.

Methodology

First, a framework is designed for identifying the instances when e-scooter riders encounter strong vibrations. Typically, those vibration events occur at potholes, cracks, or slopes during an e-scooter trip. Taking advantage of mobile sensing technologies, vibration and GPS data are recorded. Then, the collected data is processed and vibration events can be detected. Second, the parameters used in this framework are calibrated by testing e-scooters at selected locations. The tuned model is sensitive to vibration events that capture riders’ experienced vibrations. Last, other variables such as speed are also calculated and discussed along with the experienced vibrations.

Capturing Riding Vibration

The research team adopted a mobile sensing application to record naturalistic riding data for each e-scooter trip. As shown in Figure 1a, the mobile device was attached to the e-scooter’s deck. Three-dimensional accelerometer data $X_{t}$ , $Y_{t}$ , and $Z_{t}$ were selected for evaluating vibrations for the time stamp $t$ . It should be noted that $+ X$ is identical to the e-scooter’s heading direction. The data collected by these settings are very sensitive, with a minimum time resolution of 10 ms. Figure 1b shows raw vibration data for an exemplar trip. Major changes in the measurements can be observed in $X$ and Z dimensions. Then the vibration magnitude $M_{t}$ was calculated using the following equation.

M_{t} = \sqrt{{X_{t}}^{2} + {Y_{t}}^{2} + {Z_{t}}^{2}}

(1)

Maximum value $M_{max}$ and standard deviation $M_{sd}$ were calculated for each second during the trip. The calculated $M_{max}$ was taken as the vibration magnitude for each second, which can be further mined to identify significant vibration events as shown in Figure 1e. $M_{max, i}$ and $M_{max, i - 1}$ in the X–Y scatterplot denote the vibration magnitude for the $i^{th}$ second and its previous second. The blue triangle points indicate timestamps with vibrations larger than $θ$ and greater than that of previous timestep. On the other hand, another parameter $σ$ shown in Figure 1d was used for as the threshold for $M_{sd}$ . The two parameters $θ$ and $σ$ were used for determining vibration events based on the following rules: (1) $M_{max, i} > θ$ ; and (2) $M_{sd, i} > σ$ .

Figure 1.

Framework for extracting vibration events: (a) illustration of mobile phone attached the e-scooter deck, (b) raw sensor measurements, (c) computed vibration results, (d) maximum vibration and standard deviation per second, and (e) compare virbration between current and prior timesteps.

Calibrating Vibration Events

The vibration thresholds $θ$ and $σ$ in the previous section should be calibrated and tuned to ensure that correct vibration events can be detected. Figure 2 shows three typical vibration events at selected locations. Event A is a typical pothole, whereas B and C are two types of slopes. E-scooter riders will feel remarkable vibrations when crossing those locations compared with riding on flat pavements. Both small-wheel and large-wheel e-scooters were ridden multiple times between the two locations P1 and P2. A mobile device was attached to the testing scooter and vibration data was collected. Then, vibration events were identified following the workflow as described previously. The vibration thresholds $θ$ and $σ$ were tuned so that the detected vibration events were consistent with the presence of the potholes and slopes.

Figure 2.

Calibrating vibration events with small- and large-wheel e-scooters.

Measuring Riding Velocity

Riding speed is a critical factor for e-scooter safety. When riding in regions without a speed limit, e-scooters can easily reach a hazardous speed of 15 mph in a short period of time. The e-scooter’s velocity can be tracked and calculated during a trip. In this study, the velocity in each second was calculated using

V_{i} = \frac{Dist (i - 1, i) + Dist (i, i + 1)}{2}

(2)

where $Dist (i - 1, i)$ indicates the distance between the locations of $(i - 1)^{th}$ and $i^{th}$ second.

Design of Experiments

Route Design

According to the work by Jiao and Bai (2020) ( 22 ), land use is one of the primary factors affecting the demand for e-scooters. This study is focused on the sidewalks near three typical built environments: (1) sidewalks in neighborhoods; (2) sidewalks near university campus; and (3) sidewalks on commercial streets. The three selected sidewalk routes are shown in Figure 3a, and Figure 3b contains the photos of their pavements. It can be observed that neighborhood sidewalks have comparably narrow lanes, where large cracks with weeds can be frequently observed. Campus sidewalks are wider and smoother than the neighborhood sidewalks. Both of them have concrete pavement. The pavement of the commercial streets primarily consists of bricks, where different vibration patterns are expected.

Figure 3.

Selected sidewalks for conducting experimental tests: (a) test routes and (b) examples of given conditions.

Scenario Design

Two members from the research team acted as riders and conducted the relevant experiments. It should be noted that this study does not focus on the impact of human factors on riding risk. Instead, it focuses on quantifying vibrations during e-scooter trips. Arguably, the vibration patterns should be similar between the two under each scenario. This assumption was tested with statistical methods, described in later sections. Two types of e-scooter distributed by a major service provider in the City of Norfolk were tested: (1) type A has small wheels (diameter: 8 in.); and (2) type B has large wheels (diameter: 10 in.). Both types of e-scooter were ridden on the three designed routes. Meanwhile, the distributions of detected vibration events were statistically compared between the two types of e-scooter.

Calculating Vibration Indexes

After applying the calibrated model, vibration events were identified in each scenario. Then, sets of vibration indexes were calculated: (1) number of events; (2) e-scooter trip length; and (3) vibration density. The vibration density was equal to the number of events divided by the trip length, which describes the density of encountered vibration events during an e-scooter ride. The higher the vibration density, the less suitable that particular type of facility was for riding e-scooters on.

Mann Whitney Wilcoxon Rank Sum Test

The Mann Whitney Wilcoxon (MWW) rank sum test has been widely used for comparing whether two sample values are from the same population ( 23 ). The MWW test is given by

U_{t} = \frac{1}{n_{1}} \frac{1}{n_{0}} \sum_{i_{1} = 1}^{n_{1}} \sum_{i_{0} = 1}^{n_{0}} I {y_{i_{1} 1} \leq y_{i_{0} 0}}

(3)

where $y_{i_{i} k}$ denotes the observed version of the potential outcome corresponding to the treatment received ( $k = 0, 1$ ). $I (A)$ is an indicator, with $I (A) = 1$ if $A$ is true and 0, otherwise.

In this study, the detected vibration events were aggregated every 0.1 mi along the route. Then, the MWW rank sum test was applied for testing whether the performances between two riders were similar with the same experimental configuration, and if the performances were similar between the two types of e-scooter.

Experimental Results and Analysis

Vibration Event Calibration Results

As illustrated in Figure 2, a large-wheel e-scooter and a small-wheel e-scooter were ridden back and forth 10 times at selected locations. After collecting and processing vibration data, the maximum value and standard deviation of vibration magnitude are shown in Figure 4. Then, the parameters for identifying vibration events were calibrated. As a result, $θ = 3$ and $σ = 0.6$ were derived based on the following two rules: (1) no vibration event can be detected on “Flat” scenarios; (2) significant vibration events can be observed at typical potholes or slopes. There are several significant patterns that can be observed from Figure 4. First, at the location for event A, the frequencies of the detected vibration events are similar between the e-scooters with the two types of wheels. On the other hand, intensive vibration events were detected for the e-scooter with small wheels whereas fewer vibration events can be detected for the large-wheel e-scooter at locations for events B and C. Therefore, both small-wheel and large-wheel e-scooters are sensitive to vibration events at potholes like event A. Meanwhile, only small-wheel e-scooters are sensitive to vibration events at slopes like events B and C. Based on the number of vibration events, the large-wheel e-scooter performed better when riding across slopes with fewer experiences of vibrations.

Figure 4.

Vibration calibration for typical events: (a) flat surface with small wheel, (b) flat surface with large wheel, (c) event A with small wheel, (d) event A with large wheel, (e) event B with small wheel, (f) event B with large wheel, (g) event C with small wheel, and (h) event C with large wheel.

Based on the calibration, Figure 5 shows the maps for detected vibration events on selected testing routes. Because of the possible disturbance of GPS signals, the events are located within a 10 m range of the potholes/slopes. For each testing scenario, a scatter plot was drawn with the X-axis indicating event speed and Y-axis indicating calculated vibration. The correlation values between the X and Y axes were close to zero for all testing scenarios. The weak correlation suggests that the speed and vibration variables are not highly related to each other.

Figure 5.

Vibration events on selected test routes: (a) location A, (b) location B, and (c) location C.

Comparative Analyses for E-Scooters with Different Wheel Sizes

After applying the calibrated model with $θ = 3$ and $σ = 0.6$ , vibration events were detected, and the corresponding indexes are shown in Table 1. In general, riders with large-wheel e-scooters encountered fewer vibrations compared with those with small-wheel e-scooters. Large-wheel e-scooters had smaller vibration densities on all three selected routes. Specifically, the campus sidewalks are most comfortable for riding small-wheel e-scooters; and riders with large-wheel e-scooters feel most comfortable on commercial street sidewalks. One interesting observation is that the two types of e-scooter have distinctive performances when riding on the commercial street where bricks are the primary pavement. It can be inferred that small-wheel e-scooters are sensitive to the bricks, generating strong vibration patterns because of cracks, whereas large-wheel e-scooters are not affected. This observation will be further explored in later sections.

Table 1.

Vibration Indexes for Small-Wheel and Large-Wheel Scenarios

E-scooter type	Facility	Vibration event	Length (mi)	Vibration density
Small wheel	Neighborhood sidewalk	44	0.48	91.7 events/mile
Small wheel	Campus sidewalk	64	1.68	38.1 events/mile
Small wheel	Commercial street sidewalk	43	0.57	75.4 events/mile
Large wheel	Neighborhood sidewalk	27	0.48	56.3 events/mile
Large wheel	Campus sidewalk	22	1.68	13.1 events/mile
Large wheel	Commercial street sidewalk	3	0.57	5.3 events/mile

Figure 6 shows the vibrations in each second and the detected vibration events. It provides a straightforward illustration of the distributions of the events. Moreover, those spatial data can be used for developing an alerting system for e-scooter riders. Determined by the type of e-scooters, the riders can be notified when vibration events are near their riding paths.

Figure 6.

Comparative analyses of vibration for e-scooters with small and large wheels: (a) samll wheel on route 1, (b) large wheel on route 1, (c) samll wheel on route 2, (d) large wheel on route 2, (e) samll wheel on route 3, and (f) large wheel on route 3.

MWW Rank Sum Testing Results

Two-Rider Comparison

Table 2 shows the results of MWW tests by pairing two e-scooter riders in the six designed scenarios (three routes by two e-scooter types). All p-values presented are larger than 0.1. Therefore, the null hypothesis on both riders experiencing the same vibrations failed to be rejected. The testing results indicate that the differences between the two e-scooter riders can be ignored.

Table 2.

Mann Whitney Wilcoxon Rank Sum Test Between Two Riders

Wheel size	Route	p-value
Small wheel	1	0.7827
Large wheel	1	0.1666
Small wheel	2	0.6874
Large wheel	2	0.8858
Small wheel	3	0.8079
Large wheel	3	0.6733

Large-Wheel and Small-Wheel Comparison

Similarly, the MWW tests were applied to compare whether there is a statistical difference between the trips with small-wheel e-scooters and those with larger wheels. The p-values for the three selected routes are 0.0439, 0.0008, and 0.0043, respectively. The small p-values indicate that there is significant difference in detected vibration events when using e-scooters with different wheel sizes.

Discussion

The above comparative analysis indicates that different facilities have different impacts on the riding experience. Riding on sidewalks with poor pavement conditions can be a risk because of the increased vibration events. E-scooters may bump up at some unidentified potholes shown in Figure 7a. Facility maintenance engineers are expected to find such hidden hazards promptly and efficiently and provide necessary treatments. Before repair, such cracks can be marked to warn users, as in Figure 7b. Traffic cones shown in Figure 7c can be placed at a hazardous crack and temporarily protect e-scooter riders before any other engineering treatment is implemented. Even for repaired sites, if they were not as flat as the new one, it would be helpful to mark them with additional visual clue to raise the awareness of e-scooter riders (e.g., Figure 7d). Nevertheless, as many factors can lead to the degrading of sidewalk quality, timely identification of potholes and cracks in large areas can be challenging. As a mutual supporting solution, the mobile sensing procedures developed in this study can be used for gathering data and locating sites with issues. If such mobile sensing systems were deployed on existing e-scooter fleets, it will be beneficial to crowdsource the notable vibration events and quickly map the facilities with issues. The mapping results can be used to guide maintenance engineers to allocate resources for timely treatment of higher priority sites.

Figure 7.

Sidewalk cracks and treatments for safety improvement: (a) unidentified, (b) identified crack, (c) block hazard crack, and (d) fixed crack.

Conclusions

The rapid rise of e-scooter programs provides riders a new option for short-distance travel. By sharing facilities with other transport modes, no dedicated road resources have been allocated to e-scooters. Consequently, many cities do not have clear guidance on where and how to ride e-scooters safely. Besides the complex naturalistic environments, e-scooter riders can feel strong vibrations on sidewalks with low-quality pavement conditions. Urgent efforts are necessary to support safer e-scooter operations by understanding riding behaviors. As emphasized by Ma et al. ( 8 ), e-scooter riders encounter strong vibrations because of the small wheel size compared with bicycles with relatively larger wheels. Meanwhile, the vibration patterns vary across different types of pavements (e.g., concrete or asphalt). However, no sufficient work has explored the vibrations of e-scooters with different wheel sizes. To shorten this gap, this study used mobile sensing technology to collect vibration data for three designed routes. With both statistical analysis and in-depth data mining, it was found that e-scooters with larger wheels have significantly improved comfort over those with small wheels, campus sidewalks with great pavement conditions offer a good riding condition for both types of e-scooter, and large-wheel e-scooters worked better on commercial streets with brick pavement. Thus, it is suggested that the deployment of e-scooters should consider both the riding infrastructure and the vehicle characteristics to provide users with a better riding experience.

Future Scope

This study has developed procedures to quantitively assess the performances of e-scooters with different wheel sizes. The experiments were conducted on typical routes frequently used by e-scooters. It has revealed that e-scooters with larger wheels can better alleviate vibrations than those with small wheels. The vibration detection workflow with mobile sensing technologies is also portable and extendable for additional related studies. For example, e-scooters with vibration sensors installed can be used for evaluating pavement conditions. Vibration event density (VED) can be calculated as an important indicator. The lower the VED values, the better the road conditions. When the VED values reach a pre-defined threshold, necessary maintenance or on-site inspection can be considered.

Besides the contributions and possible future applications, this study is limited in several aspects. First, besides the wheel size, the two types of e-scooter involved in this study also have different structures for damping springs. Therefore, the wheel size may not be the only reason for the experiment results. Further experiments are necessary if the damping spring structures were considered. Second, it should be noted that because of the availability of the e-scooters in the study area, only two types of e-scooter were tested. As more models of e-scooters emerge (e.g., e-scooters with portable batteries and e-scooters with different front and rear wheels), extension of the experimental study to additional types of e-scooters will further strengthen the understanding of their possible riding issues. Finally, the proposed vibration evaluation method requires researchers to attach mobile devices to e-scooters. This method is difficult to scale without vendors’ support. Thus, collaboration with e-scooter vendors is encouraged to deploy such sensors on e-scooters for a broader understanding of riding behaviors as well as providing data for timely assessment of facility conditions.

Footnotes

Acknowledgements

The study design, experiments, and data collection work have been performed while the first author was at Old Dominion University. We thank the previous undergraduate students A. Mayhue and R. Sun for their help in this research project. The authors appreciate the reviewers for their great comments that helped improve the paper.

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: H. Yang, Q. Ma; data collection: Q. Ma, Z. Yan, H. Yang; analysis and interpretation of results: Q. Ma, H. Yang, Z. Yan; draft manuscript preparation: Q. Ma, H. Yang, Z. Yan. 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 was partially supported by a Program for Undergraduate Research and Scholarship (PURS 2020) grant from the Office of Research and Perry Honors College at Old Dominion University, Norfolk, Virginia, USA.

ORCID iDs

Qingyu Ma

Hong Yang

Data Accessibility Statement

The data used in this study were obtained based on individual riding tests. For privacy reasons, the data will not be publicly released. However, the data can be made available based on rational request and necessary data desensitization.

The contents of this paper only reflect views of the authors who are responsible for the facts and do not represent any official views of any sponsoring organizations or agencies mentioned in this paper.

References

DDOT. Shared Fleet Device Program. https://ddot.dc.gov/node/1323886. Accessed February 28, 2023.

FHWA. National Household Travel Survey. https://nhts.ornl.gov/vehicle-trips. Accessed February 28, 2023.

Yang

Xie

Examining Municipal Guidelines for Users of Shared E-Scooters in the United States. Transportation Research Part D: Transport and Environment, Vol. 92, 2021, p. 102710.

de Bortoli

Christoforou

Consequential LCA for Territorial and Multimodal Transportation Policies: Method and Application to the Free-floating E-Scooter Disruption in Paris. Journal of Cleaner Production, Vol. 273, 2020, p. 122898.

Yang

Wang

Cai

Xie

Yang

Safety of Micro-mobility: Analysis of E-Scooter Crashes by Mining News Reports. Accident Analysis & Prevention, Vol. 143, 2020, p. 105608.

Posirisuk

Baker

Ghajari

Computational Prediction of Head-Ground Impact Kinematics in E-Scooter Falls. Accident Analysis & Prevention, Vol. 167, 2022, p. 106567.

Currans

K. M.

Iroz-Elardo

Ewing

Choi

D.-a.

Scooting to a New Era in Active Transportation: Examining the Use and Safety of E-scooters. Report No. #NITC-RR-1281. National Institute for Transportation and Communities, 2022.

Yang

Mayhue

Sun

Huang

E-scooter Safety: The Riding Risk Analysis Based on Mobile Sensing Data. Accident Analysis & Prevention, Vol. 151, 2021, p. 105954.

Toll

Lime Unveils New Electric Scooter, Fuels the E-Scooter Arms Race. https://electrek.co/2018/10/19/lime-s-generation-3-electric-scooter/. Accessed August 1, 2021.

10.

Xin

Yang

Xie

Connecting Metros With Shared Electric Scooters: Comparisons With Shared Bikes and Taxis. Transportation Research Part D: Transport and Environment, Vol. 109, 2022, p. 103376.

11.

Shichman

Shaked

Factor

Weiss-Meilik

Khoury

Emergency Department Electric Scooter Injuries After the Introduction of Shared E-scooter Services: A Retrospective Review of 3,331 Cases. World Journal of Emergency Medicine, Vol. 13, No. 1, 2022, p. 5.

12.

Azimian

Jiao

Modeling Factors Contributing to Dockless E-scooter Injury Accidents in Austin, Texas. Traffic Injury Prevention, Vol. 23, No. 2, 2022, pp. 107–111.

13.

Gössling

Integrating E-scooters in Urban Transportation: Problems, Policies, and the Prospect of System Change. Transportation Research Part D: Transport and Environment, Vol. 79, 2020, p. 102230.

14.

Tuncer

Laurier

Brown

Licoppe

Notes on the Practices and Appearances of E-Scooter Users in Public Space. Journal of Transport Geography, Vol. 85, 2020, p. 102702.

15.

Fitt

Curl

E-scooter Use in New Zealand: Insights Around Some Frequently Asked Questions. University of Canterbury: Christchurch, New Zealand, 2019.

16.

Sikka

Vila

Stratton

Ghassemi

Pourmand

Sharing the Sidewalk: A Case of E-scooter Related Pedestrian Injury. The American Journal of Emergency Medicine, Vol. 37, No. 9, 2019, pp. 1807.e1805–1807.e1807.

17.

Cano-Moreno

J. D.

Marcos

M. I.

Haro

F. B.

D’Amato

Juanes

J. A.

Heras

E. S.

Methodology for the Study of the Influence of E-scooter Vibrations on Human Health and Comfort. Proc., Seventh International Conference on Technological Ecosystems for Enhancing Multiculturality, León, Spain, 2019, pp. 445–451.

18.

APH. Dockless Electric Scooter-Related Injuries Study. https://www.austintexas.gov/sites/default/files/files/Health/Epidemiology/APH_Dockless_Electric_Scooter_Study_5-2-19.pdf. Accessed June 14, 2021.

19.

Zou

Younes

Erdoğan

Exploratory Analysis of Real-Time E-scooter Trip Data in Washington, DC. Transportation Research Record: Journal of the Transportation Research Board, 2020. 2674(8): 285–299.

20.

Paudel

Yap

F. F.

Bastola

A. K.

Safety Assessment of Personal Mobility Devices With Different Wheel Size Based on Their Dynamic Stability Performance. International Journal of Sustainable Design, Vol. 3, No. 4, 2020, pp. 227–250.

21.

Iftakhar

Are E-Scooters Suitable for UK Roads? The Pothole Test for Safe Scooter Design. https://swiftyscooters.com/blogs/journal/e-scooter-safety-and-design. Accessed November 30, 2021.

22.

Jiao

Bai

Understanding the Shared E-scooter Travels in Austin, TX. International Journal of Geo-Information, Vol. 9, p. 135.

23.

Han

Chen

Causal Inference for Mann–Whitney–Wilcoxon Rank Sum and Other Nonparametric Statistics. Statistics in Medicine, Vol. 33, No. 8, 2014, pp. 1261–1271.