This session includes papers that describe research about collecting and analyzing pedestrian and bicyclist data.
A New Tool for Pedestrian Observation: Reliability and Efficiency of Small Unmanned Aircraft Systems
Keunhyun Park, Utah State UniversityShow Abstract
Reid H. Ewing, University of Utah
The monitoring of pedestrian activity is challenging, primarily because its traffic levels are typically lower and more variable than motorized vehicles. Compared with other on-the-ground observation tools, the small Unmanned Aircraft Systems (UAS) could be suitable for counting and mapping pedestrians in a reliable and efficient way. Thus, this study establishes and tests a new method of pedestrian observation using UAS. The results show that the UAS observations demonstrate high levels of inter-rater reliability (ICC = 0.99) and equivalence reliability (Cronbach’s a = 0.97 (with on-the-ground counts); 0.73 (with Google Street View)). Practical implications of the new tool are discussed.
Bicycle and Pedestrian Count Programs: A Scan of Current U.S. Practice
Peter Ohlms, Virginia Department of TransportationShow Abstract
Lance Dougald, Virginia Department of Transportation
Hannah E. MacKnight, Virginia Transportation Research Council
As bicycling and walking have become more integrated into transportation agencies’ processes of planning, design, and operations, some state, regional, and local agencies have established nonmotorized data collection programs of varying scopes and with varying methods. The purpose of this study was to identify ways to plan and implement a nonmotorized count program, and the scope included reviewing existing national-level guidance and examples from other state DOTs to determine the most effective ways of implementing such a program. Study tasks included synthesizing the literature to obtain relevant information regarding nonmotorized travel monitoring programs, practices, and technologies, as well as obtaining information from representatives of three states through interviews of public agency staff and/or researchers involved in each state’s program. The study found a large volume of recent research on the topic of nonmotorized travel monitoring. The study concluded that the state of the practice in nonmotorized travel monitoring has evolved and expanded in recent years; that many commercially available counting technologies exist and have been evaluated; that the practice of nonmotorized travel monitoring, as with motorized travel monitoring, has several aspects beyond purchase and installation of automatic count equipment; and that several states are developing nonmotorized count programs and have begun putting their data to use. The findings provide a foundational resource for state DOTs that are considering developing state-level counting programs.
Development of Instrumented Bicycle and Mobile Applications to Perform Cloud-Based Pavement Condition Management for Bike Roads
Chun-Hsing Ho, Northern Arizona UniversityShow Abstract
Peijie Qiu, Northern Arizona University
Siwei Wen, Northern Arizona University
Xilun Liu, Northern Arizona University
Matt Snyder, Northern Arizona University
Kyle Winfree, Northern Arizona University
The paper introduces the instrumented bicycle sensing technology developed by the research team, presents a mobile app named, “Instrumented Bike” and a Google cloud website to demonstrate how the instrumented bicycle advances cycling activities in detecting potential hazards and displays the locations of georeferenced hazards on bike routes. The instrumented bicycle senses road conditions through the sensor logger along with the mobile app. The sensor logger consists of microprocessor, accelerometers, Global Positioning System (GPS) unit, battery, computer board (Arduino board), and Wi-Fi device, paired with a smartphone via the Instrumented Bike mobile app. All real-time data is immediately transferred to the Mobile app and the cloud server where all data is analyzed using computing algorithms developed by the research team. Potential hazards are identified, georeferenced, and displayed in Google maps to allow bicyclists (road users) and governments/authorities (road managers) to review prior to their cycling or decision-making for day-to-day maintenance. The paper concludes that the instrumented bicycle sensing technology presented in the paper allows cycling activities to effectively and immediately reflect and communicate the real-time road conditions to the bicyclists/governments who share/manage the bike routes, market and promote bike mobility, and recruit more participants in cycling data collection.
Multi-Directional Pedestrian Detection and Flow Monitoring from Traffic Videos
RITWIKA CHOWDHURY, IIT KharagpurShow Abstract
Kinjal Bhattacharyya, Indian Institute of Technology Kharagpur
Sudipta Mukhopadhyay, Indian Institute of Technology Kharagpur
Prof. Bhargab Maitra, Indian Institute of Technology, Kharagpur
Rapid urbanization and increasing road traffic has paved the way for the design of an efficient traffic management system. In this paper, we present a computer vision and queueing theory based technique to detect pedestrians and compute key macroscopic statistics of a pedestrian traffic. We performed the task of pedestrian detection in a specified region of interest in a traffic video using an Aggregated Channel Feature based detector The detector was pretrained using INRIA dataset and fine-trained with Varanasi dataset to eliminate certain false positives appearing in mixed traffic scenarios such as detecting bike riders as pedestrians. This allowed the detector to work with, however not limiting its scope to, heterogeneous traffic prevalent on roads of emerging countries. We further employed Little’s Law from the realm of Queueing theory to estimate pedestrian flow rate. This novel method eliminates the need for using existing resource intensive tracking based approaches for finding pedestrian flow. The whole pipeline is computationally efficient while not compromising on performance metrics. Our proposed method can further be implemented in-situ at the traffic monitoring camera sites to realise a distributed traffic management system. Such a system will be robust, scalable, autonomous and can provide real-time assistance for pedestrian control.
Processing Cycling Risk Under Different Elicitation Methods: Comparing 2D and 3D in Virtual Reality Choice Environments
Martyna Bogacz, University of LeedsShow Abstract
Chiara Calastri, University of Leeds
Charisma Choudhury, University of Leeds
Stephane Hess, University of Leeds
Alex Erath, Future Cities Laboratory
Michael Van Eggermond, Future Cities Laboratory
Faisal Mushtaq, University of Leeds
The aim of this study is to provide a better understanding of cyclists’ risk perception in different scenarios under different elicitation methods. In particular, 2D computer-based videos and 3D virtual reality simulations of road situations are contrasted. We collect data on cyclists’ behavioural responses in risky conditions and their stated responses on propensity to cycle and risk perception. Electroencephalography (EEG) is used to gain insight into the temporal sequence of cortical risk processing, which gives a better understanding of neural mechanisms underlying choices. In addition, this study provides the validation of virtual reality as a tool for risk preference elicitation. Our results are in line with expectations: they show behavioural responses in line with the stimuli of the scenarios and an effect of the elicitation method, e.g. the perception of the riskiest elements seem to be exacerbated in 3D. Overall, we show that the 3D presentation method has an impact on the neural processing of risk and not only it changes the way people perceive risk but also their behaviour. The findings provide useful insights about data collection in the context of cycling behaviour and beyond.
A Pedestrian-Oriented Framework for Measuring Area-Wide Pedestrian Activity
Steven Gehrke, Metropolitan Area Planning CouncilShow Abstract
Peter James, Harvard Medical School
Halley Reeves, Massachusetts Department of Public Health
Sharon Ron, Metropolitan Area Planning Council
Timothy G. Reardon, Metropolitan Area Planning Council
Barry Keppard, Metropolitan Area Planning Council
W.W. Sanouri Ursprung, Massachusetts Department of Public Health
Research continuously suggests the built environment provides opportunities as well as barriers to active travel and physical activity. Individuals residing in densely-populated neighborhoods with a mix of land uses tend to walk more. While insights into how neighborhood-level built environment features impact individual-level behavior continue to inform urban policies, research centered on measuring active travel at a neighborhood scale is needed to estimate the population-level impact of policy, systems, or environmental changes on transportation-related physical activity. To date, a nascent body of research has sought to create the requisite tools for measuring area-wide levels of active travel, albeit by applying existing vehicle-based methods. Our study advances current practice by introducing a pedestrian-oriented approach to classifying streets based on a measure of local destination accessibility along a given street segment, or its network utility, and pedestrian count data collected from multiple randomly-selected sites in four neighborhoods across Massachusetts. As a proof of concept, in one study area, data collected with automated counters using our pedestrian-based street stratification method were expanded in order to create area-wide estimates of seasonal average daily pedestrian counts that were in turn used to estimate average daily pedestrian miles traveled. An area-wide pedestrian activity estimate, which resulted from a method that can establish pre/post-intervention statistics of neighborhood-level pedestrian travel and physical activity needed to inform evidence-based policies.
Comparative Analysis of User Behavior of Dock-Based and Dockless Bikeshare and Scootershare in Washington, D.C., Metropolitan Area
Kiana Roshan Zamir, University of Maryland, College ParkShow Abstract
Iryna Bondarenko, University of Maryland, College Park
Sean Thomas Burnett, University of Michigan, Ann Arbor
Stefanie Brodie, District Department of Transportation
Kim Lucas, District Department of Transportation
In 2017, dockless bikeshare systems were introduced in the United States and dockless scootershare followed in early 2018. These new mobility options complement existing station-based bikeshare systems, which are bound to static origin and destination points at docking stations. The users of dockless bikeshare, dockless scootershare, and dock-based bikeshare have different travel behavior and this research presents a comparative analysis of users’ behavior of these three modes in Washington, DC. It uses logistic regression and random forest modeling to delineate between “member” behavior, which aligns most closely with commuter behavior, and “casual” behavior that represents more recreational behavior. The results show scooter riders demonstrated strong commuter characteristics (i.e. short trip duration and start morning trips from mixed-use neighborhoods) that are similar to Capital Bikeshare members. Dockless bikeshare’s users showed many casual riders’ characteristics; however, the models classified most of these trips as “members trips.” Overall, this study provides evidence that dockless bicycles and scooters have a range of uses and complement the existing bikeshare system; therefore, they can make positive contributions to urban multi-model infrastructure. Keywords: User Behavior, Bikeshare, Dockless Bicycles, Dockless Scooters, Classification, Casual Riders, Registered Members, Capital Bikeshare
A Methodology to Incorporate Scenicness into Revealed Preference Pedestrian Route Choice Modeling from Smartphone Data
Parham Hamouni, Concordia UniversityShow Abstract
Ciprian Alecsandru, Concordia University
Zachary Patterson, Concordia University
This study builds on a growing body of research that seeks to understand the factors influencing pedestrian route choice. Better understanding of these factors can help in the promotion of walkability. Pedestrian route choice is derived from the data of a smartphone travel survey conducted in the fall of 2016 in Montreal, Canada. Trip route characteristics are obtained by matching smartphone traces with the underlying road network with the Open Source Routing Machine (OSRM). In addition to traditional geometric features, itineraries were associated with a proposed scenicness indicator derived from Google Streetview Images, the Places365 Convolutional Neural Network and a scenicness Elastic Net model. Finally, establishment types from the Google Places API were also associated to itineraries for use in route choice modeling. A path-size multinomial logit model is used to assess the utility of the explanatory variables. Additionally, to improve prediction accuracy, a set of supervised learning classification techniques, including decision tree, random forest and gradient boosting tree were examined. The paper demonstrates the feasibility of the approach to incorporate scenicness into route choice models. Moreover, the analysis of the results shows that, as suggested in research based on stated preference, variation in scenery has a significant impact on pedestrian route choice. Furthermore, the machine learning classification techniques show improvements in route choice prediction when compared to the discrete choice modeling framework.
A Pedestrian Exposure Model for the California State Highway System
Julia Griswold, University of California, BerkeleyShow Abstract
Aditya Medury, Safe Transportation Research and Education Center
Robert Schneider, University of Wisconsin, Milwaukee
Dave Amos, University of California, Berkeley
Offer Grembek, University of California, Berkeley
For this study, we developed one of the first statewide pedestrian exposure models, using log-linear regression to estimate annual pedestrian crossing volumes at intersections on the California State Highway System. We compiled a database of more than 1,200 count locations, one of the largest ever used to create a pedestrian volume mode. We initially evaluated 75 explanatory variables for the model. The final model is based on the three land use variables (employment density, population density, number of schools), four roadway network variables (number of street segments, intersections with principal arterial and minor arterial roadways, and four-way intersections), and American Community Survey journey-to-work walk mode share that are readily-available or fairly easy to create using basic GIS analysis. The resulting pedestrian volume model was used to estimate annual crossing volumes at more than 12,000 intersections along the California State Highway System. This is one of the first statewide pedestrian volume models, and California may represent the largest jurisdiction to date to adopt a single, system-wide pedestrian volume model.
Evaluating OpenStreetMap's Performance Potential for Level of Traffic Stress Analysis
David Wasserman, Fehr & PeersShow Abstract
Alex Rixey, Fehr & Peers
Xinyi (Elynor) Zhou, Fehr & Peers
Drew Levitt, Fehr & Peers
Matt Benjamin, Fehr & Peers
Increasingly, metropolitan areas are prioritizing growth in the share of trips taken by bicycle to improve health outcomes, transportation affordability, and environmental performance of the transport system. Evidence is building that network quality is an important determinant in bicycle commuting and route choice. One prominent metric of facility attractiveness is Bicycle Level of Traffic Stress (LTS). In tandem, OpenStreetMap (OSM) is becoming an important source of network data for routing and for generating measures of multimodal accessibility. While there are studies that examine the completeness of OSM tags and utilize OSM data to compute LTS on networks, none of them examine the accuracy of these analyses. The goal of this paper is to evaluate the accuracy of OSM-derived LTS predictions and offer quality assurance strategies to reduce inaccurate predictions. This study compares OSM-derived LTS predictions with ground-truthed LTS scores created by Montgomery County. We find that OSM-derived LTS networks provide comparable results to the ground-truthed data. The OSM-derived LTS scores correctly identified 89.9% of the length of the network as either high (LTS 3 or 4) or low stress (LTS 1 or 2). However, this study demonstrates there is a higher potential for error within certain street typologies and urban contexts, and that low stress accessibility calculations can be very sensitive to even a small number of incorrectly classified segments. Finally, we suggest practices to improve the quality of OSM-derived LTS predictions and low-stress accessibility calculations.
The Relationship Between Macroscopic Cycling Traffic Parameters on Cycle Tracks
David Beitel, McGill UniversityShow Abstract
Miguel Domínguez-Michelen, McGill University
Luis Fernando Miranda-Moreno, McGill University
Cycling travel time must be competitive with other modes of transportation such as private vehicle or public transportation in order for cycling mode share to grow substantially. It is essential for transportation agencies to have the tools to quantify non-motorized transportation demand and knowledge of infrastructure needs for planning and design purposes. This study combines cyclist GPS trip data, from the Mon RésoVélo smartphone application, bicycle count data from an automated bicycle counter, and weather data, obtained from Environment Canada, to model the impact on average bicycle speed of bicycle flow, stopping frequency and weather. The test site is a bi-directional cycle track in Montreal. In all, 289 cleaned and validated bicycle trips were analysed over five blocks: bicycle speed was extracted and related to bicycle flow data from an automated bicycle counter along the same segment of study. Cyclist flow was determined to be a statistically significant predictor (99% confidence level) of average cycling speed. As cyclist flow increases by 100 cyclists/15 minutes, average cyclist speed decreases by 0.5 m/s. The temperature, wind speed and the proportion of time stopped were also found to be statistically significant predictors of average bicycle speed. A linear regression model to estimate bicycle speed was developed, with R2 = 0.407.
Prediction of Transfer Probability of Traffic Flow for Shared Bikes Based on Hybrid Optimization and Deep Learning Algorithms
Wenwen Tu, Southwest Jiaotong UniversityShow Abstract
Hengyi Liu, University of Waterloo
Shared Bikes refers to the bike-sharing services across the residential, the commercial, and the public service areas. Predicting the destinations of the cycling trips and the traffic flow transfer of shared bikes between the traffic zones can improve the dispatching efficiency of shared bikes, recommendations of possible locations for the users, and improve the navigation efficiency. This paper proposes a stacked Restricted Boltzmann Machine (RBM)-Support Vector Regression (SVR) deep learning algorithm for predicting the transfer probability of traffic flow of Shared Bikes. To further improve the accuracy of prediction, the parameters in the stacked RBM-SVR algorithm were optimized by proposing a hybrid optimization algorithm named DEGWO, which obtained by merging the Differential Evolution (DE) algorithm and the Grey Wolf Optimization (GWO) algorithm. In an experimental case, the destinations of the cycling trips and the probability of traffic flow transfer for shared bikes between traffic zones were predicted by computing 2.46 million trajectory points recorded by shared bikes in Beijing. By making comparisons, it revealed that the stacked RBM-SVR algorithm, with the help of the DEGWO algorithm, outperformed the other two conventionals. As a result, the prediction results were much more accurate.
Stated Preference Analysis of Bicycle Sharing Systems in Chinese Cities
Yi Zhu, Shanghai University of Economics and FinanceShow Abstract
Zhao Guyue, Shanghai Jiao Tong University
Bicycles used to play an important role in urban transportation of Chinese cities decades ago, but it had been gradually replaced by private cars, metro, bus and some other modes, owning to the fast-growing mobility demand as a result of urban expansion and motorization. However, in recent years, with the development of ICT technologies and initiative of sharing economy, bike sharing systems (BSSs) have been implemented extensively in Chinese cities. Usage patterns can be revealed via system generated data, yet little is known about users’ attitudes and preferences of the systems. In this study, we draw on two surveys conducted in Guangzhou and Beijing on the attitudes of travelers on BSSs to estimate the effect of demographic factors, bicycle ownership and trip level factors on the willingness and potential frequency of BSS usage. In addition, latent class models are built to analyze different aspects of systems concerned by different types of urban travelers in light of their stated requirements for the systems. It is found that respondents’ age, occupation, income, mode combination, and proximity of origin or destination to docking station, etc. have an influence on the willingness and frequency of using BSSs. Also, respondents generally value the features such as proximity of docking stations to trip ends, safety to ride, and appropriate level of fare. But different latent classes show a different preference to other features of BSSs. According to model results, proposals are given for the improvement of existing systems in Chinese cities.
Large-Scale Bicycle Flow Experiment: Setup and Implementation
Alexandra Gavriilidou, Delft University of TechnologyShow Abstract
Maria Wierbos, Delft University of Technology
Winnie Daamen, Delft University of Technology
Yufei Yuan, Delft University of Technology
Victor Knoop, Delft University of Technology
Serge Hoogendoorn, Delft University of Technology
Cycling research at the operational behavioral level is limited, mainly due to the lack of empirical data. In order to overcome this data shortage, we performed a controlled large-scale cycling experiment in the Netherlands. In this paper we describe the methodology for setting up and implementing such an experiment, from the motivation of its design using a conceptual model describing cyclist behavior to adjustments that were required during the experiment. The main contribution of this paper is, therefore, to be used as a guide in future experimental data collections. Moreover, we present the characteristics of the participants and their bicycles, and provide a qualitative description of phenomena observed during the experiment. Finally, we elaborate on the potential that the collected dataset holds for future research into understanding and modeling operational cycling behavior.
Innovative Applications of Dockless Bikeshare Data: Washington, D.C., and Montgomery County, Maryland
Alexandra Frackelton, Toole Design GroupShow Abstract
Frank Proulx, Toole Design Group, LLC
Recent research, practice, and federal policy highlight the need for data-driven approaches to planning analysis and decision-making. Further, GPS-based and smartphone applications present new tools to supplement traditional approaches to collecting bicycle ridership and travel behavior data. Although there is a growing body of research employing bikesharing data for planning and analysis purposes, emerging technologies such as dockless bikeshare present an opportunity for refined, disaggregated analysis. This paper demonstrates two case study examples of leveraging dockless bikeshare data for planning applications in the Washington, DC metropolitan area. Specifically, trip data were utilized to estimate bicycle parking demand in combination with other data sources in Washington, DC, while an app-based field survey was conducted in Montgomery, County, Maryland, to inform pilot program policy recommendations. These case study results indicate that dockless bikeshare data reveals patterns about bicycle presence, condition, and travel behavior. Limitations of this analysis suggest that the utility of this data and approaches will increase as data quality and availability improve, based on local data sharing and reporting requirements.
Minimizing AADNT Estimation Errors: How Many Counters Are Needed per Factor Group
Krista Nordback, UNC Highway Safety Research CenterShow Abstract
Sirisha Kothuri, Portland State University
Dylan Johnstone, Toole Design Group
Greg Lindsey, University of Minnesota, Twin Cities
Sherry Ryan, San Diego State University
Jeremy Raw, Federal Highway Administration (FHWA)
Accurate bicycle and pedestrian volume estimates inform safety studies, monitoring trends and infrastructure improvements. FHWA’s Traffic Monitoring Guide advises current practice for nonmotorized traffic estimation. While methodologies have been developed to minimize error in estimation of Annual Average Daily Nonmotorized Traffic (AADNT), challenges persist. This study provides new guidance for nonmotorized monitoring and volume estimation. Using continuous count data from 102 sites across six cities, the findings confirm that mean average percent error (MAPE) in estimated AADNT is minimized when seven-day short duration counts are collected June through September and for 24-hour counts, when data are collected Tuesdays through Thursdays (except for pedestrian-only counts). MAPE across all days (except holidays) and seasons was 22% for seven-day and 34% for 24-hour short duration counts. The magnitude of bicycle and pedestrian volumes did not significantly impact estimation errors. For factor groups larger than one, the length of short duration samples may influence AADNT accuracy estimates more than the number of counters per group, all else equal. To maximize precision of estimates of AADNT, we recommend four or more counters per factor group for bicycle and five or more for pedestrian travel monitoring. These findings provide guidance for practitioners seeking to establish or improve nonmotorized traffic monitoring programs.
Active Transportation Indicators and Establishing Baseline in Developing Country Context: A Study in Rajshahi, Bangladesh
Shaila Jamal, McMaster UniversityShow Abstract
HOSSAIN MOHIUDDIN, University of Iowa
This paper shared the findings of an active transportation (AT) study in Rajshahi, Bangladesh. The main goal was to offer an understanding of the AT situation in a developing country city context. At first, a list of AT indicators was developed based on the literature review and expert opinion survey. This list can be used to measure the AT scenario in any municipalities of Bangladesh as well as in cities of other developing countries with little or no modification. Second, the study conducted a face-to-face survey in Rajshahi which collected individuals’ socio-demographic characteristics, travel behavior, AT mode choice, and their perceptions regarding the AT scenario in their neighborhoods. An exploratory analysis of the results of the survey findings suggests that several socio-demographic characteristics are associated with AT use. For instance, it has been seen that young people are the major users of AT. Almost three-fourths of the households own at least one bicycle. As car ownership is beyond affordability, people with low income use cycle more for traveling compared to high income people. Individuals usually prefer AT mode for commute and regular grocery shopping. They tend to walk more when travel duration is less than 10 minutes. Moreover, individuals in residential areas walk more whereas individuals living in educational areas tend to cycle more. Overall, educational areas are perceived as safer and convenient areas for using active modes of transport. Many sectors such as planning, transportation, health, and education as well as non-government organizations will be benefited from this study.
Comparing the Characteristics of Bicycle Trips Among Endomondo, Google Maps, and Mapquest
Angela Schirck-Matthews, University of FloridaShow Abstract
Hartwig Hochmair, University of Florida
Dariia Strelnikova, Carinthia University of Applied Sciences
Levente Juhász, University of Florida
Numerous online trip planners, such as Google Maps and MapQuest, offer trip recommendations for cycling among other transportation modes. Whereas these platforms are widely used, little is known about the routing criteria embedded into their routing algorithms, and, more specifically, how recommended trips differ from routes travelled by cyclists. This study uses GPS tracking data from the Endomondo bicycle app to extract bicycle trips in Miami-Dade County, Florida. It compares Endomondo trip characteristics to the characteristics of bicycle trips obtained from Google Maps and MapQuest and the corresponding shortest path. Results highlight the attributes relating to road usage along a trip (e.g. percent of residential roads), trip geometry (e.g. number of turns) and surrounding land cover (e.g. percent of tree canopy) which significantly differ between the analyzed trips. These attributes should be considered in the refinement of routing algorithms to more closely resemble observed routes and hence to improve the usability of suggested trips from a cyclist’s point of view. Furthermore, the study estimates a multinomial logit model on observed Endomondo trips that provides insight into the routing behavior of Endomondo commute and sport cyclists, respectively.
A GIS-Based Relative Weight Approach for Evaluating the Station Suitability of Bikesharing System
Wendong Chen, Southeast UniversityShow Abstract
Jingxu Chen, Hong Kong Polytechnic University
Long Cheng, Ghent University
Pengfei Wang, Southeast University
In recent years, bike-sharing system (BSS) has been put into use in many cities in an effort to provide travel convenience for local residents. The location of bike stations is a pivotal issue that significantly influences the usage of BSS. A quantitative method therefore is essential to evaluate the suitability of the bike stations’ location. To this end, the primary objective of this study is to conduct a quantitative assessment of the station location of BSS in the city of Nanjing, China based upon the available bike-sharing usage data. In order to identify the distribution of bike station suitability, a Relative Weighted Approach integrating geographic information system (GIS) is proposed to examine influence factors from bike infrastructure, land use, and built environment attributes. Later, the term “composite suitability score” is utilized to separate the numerical values into five suitability classes within the study area, namely very high, high, moderate, low and very low suitability (the corresponding proportions in the case of Nanjing study area are 2.01%, 6.32%, 11.69%, 21.26% and 58.72%, respectively). This study provides a beneficial tool for transportation planners and decision makers who wish to establish effective and appropriate measures for designing and redeploying the bike-sharing system.
Estimating the Parking Demand of Free-Floating Bikesharing: A Journey Data–Based Study of Nanjing, China
Mingzhuang Hua, Southeast UniversityShow Abstract
Shujie Zheng, Southeast University
Long Cheng, Ghent University
Jingxu Chen, Hong Kong Polytechnic University
In recent years, free-floating bike sharing (FFBS) has been rapidly promoted in China, attracting numerous users and occupying much urban space. The extensive allocation and chaotic circulation of FFBS are challenges for urban transportation, and put forward higher requirements for parking facility planning. This paper estimates the parking demand of FFBS by combining the journey data of Mobike (one of the world's largest FFBS operators) in Nanjing and the Nanjing FFBS bike survey. Several clustering methods were applied to determine the virtual stations of bike aggregation. The maximum number of bikes in a day is recognized as the parking demand in each virtual station. The results show that FFBS mainly serves short-distance trips and many bikes are in inefficient operation. The K-means method turns out to observe the best clustering result for determining virtual stations, and the suitable number of stations in the built-up area of Nanjing is calculated accordingly. In addition, the demand of parking spaces of all FFBS companies in Nanjing is estimated. The research could help to propose appropriate plan for meeting the parking demand of FFBS, and have enlightening significance for promoting the healthy and sustainable development of the FFBS industry.