-
The database of authors used for the analysis comprised 842 authors from the 300 papers obtained. To analyze the authors, we set the minimum numbers of publications and citations per author to two and one, respectively. These thresholds were established such that the analysis findings of VOSviewer would be straightforward and simple to comprehend. A total of 123 among the 842 authors satisfied the thresholds, and an analysis of the findings regarding the relationships among these scholars is shown in Fig. 3.
In Fig. 3, the size of the nodes shows the number of papers published by the authors, the color of the nodes indicates the number of citations of each document, and the distance between the nodes indicates the relevance of the scholar to others. Among these authors, the top-10 authors were listed based on the total number of documents and their number of citations, as shown in Table 1.
Table 1. Top-10 authors ranked by total documents and citations.
Rank Author Documents Author Citations 1 Kockelman K. 10 Shiftan Y. 416 2 Ceder A. (Avi) 6 Kockelman K. 389 3 Chow J. 6 Hensher A. 247 4 Jenelius E. 6 Zhao J. 225 5 Axhausen K. 5 Guerra E. 219 6 Ben-Akiva M. 5 Shen Y. 146 7 Pernestal A. 5 Zhang H. 139 8 Rau A. 5 Levin M. 136 9 Wang Z. 5 Acheampong R. 130 10 Zhao J. 5 Cugurullo F. 130 As shown in Table 1, Kockelman, a scholar from the University of Texas at Austin, has published 10 papers and is the author with the most publications. The three subsequent scholars published six papers each, i.e., Ceder A. (Avi) from the University of Auckland, Chow J. from New York University, and Jenelius E. from the KTH Royal Institute of Technology. Despite authoring four documents, Shiftan Y. from the Israel Institute of Technology ranked first with 416 citations. Professor Zhao J. from MIT and his associates Shen Y. and Zhang H. et al. contributed significantly to the field of autonomous bus services, with an average of over 40 citations per paper. In addition, as shown in Fig. 3, we discovered co-authorship relationships among 20 scholars.
Analysis of source journals
-
Statistics regarding the origins of the papers showed that 80 journals published articles pertaining to autonomous bus services. Among them, 34 journals published at least two documents and were cited at least once; their connections are shown in Fig. 4. The size of the nodes in the figure shows the number of times each journal was cited, and the color of the nodes shows the average year in which the papers were published by the journal. Moreover, the lines between the nodes indicate mutual citations between journals, where the smaller the distance between two nodes, the higher the number of mutual citations.
As shown in Fig. 4, these journals are early sources of outcomes regarding autonomous bus services, e.g., Transportation Research Record, Transportation Research Part C, and Transportation. Therefore, these journals indicate a significant number of articles and citations. As shown in Table 2, Transportation Research Record, Transportation Research Part A, and Transportation Research Part C were the top-three journals, with 41, 32, and 30 papers, respectively. Moreover, the high cross-citations between Transportation Research Record and Transportation Research Part A, as well as between Transportation and Transportation Research Part A, indicate that these journals reported highly relevant outcomes. Transportation is the journal that published early outcomes in the field, although the total number of papers was not high. Notably, although only five papers were published in Transportation, the total number of citations was 338, thus rendering it the journal with the highest average citations per document. In addition, the most recent research outcomes in this field were published in Transportation Research Part E, Transportation Science, and the KSCE Journal of Civil Engineering. The scope of these journals and the increase in the number of papers published show that research pertaining to autonomous bus services has progressed from the initial stages of acceptability, feasibility, planning, and design to the more complex aspects of service optimization.
Table 2. Top-10 journals ranked by total documents and average citations.
Rank Journal Documents Journal Citations 1 Transportation Research Record 41 Transportation 67.6 2 Transportation Research Part A 32 Journal of Transport Geography 31.0 3 Transportation Research Part C 30 Transportation Research Part C 29.4 4 Sustainability 21 Transportation Research Part A 22.2 5 Journal of Advanced Transportation 12 Transportation Research Part F 21.4 6 Transportation Research Part F 10 Transport Policy 20.9 7 Transport Policy 10 Journal of Urban Planning and Development 20.0 8 Transportation Research Part D-Transport and Environment 9 International Journal of Sustainable Transportation 18.8 9 Cities 8 Transportation Research Record 15.8 10 Research in Transportation Business and Management 7 Transportation Research Part D 14.8 Analysis of keyword co-occurrence
-
Co-occurrence analysis of keywords is a method to determine the relevance of keywords to each other by observing the frequency of their simultaneous appearance. A total of 1,480 keywords from the 300 documents were analyzed in this study, among which 87 appeared more than five times simultaneously. These 87 keywords formed a network (Fig. 5) that showed the simultaneous appearance of each keyword and how they were related. The size of the nodes in the figure indicates the number of times a keyword appears, and the same-colored nodes indicate that these keywords exhibit high relevance and can be classified into a cluster.
As shown in Fig. 5, the keywords related to autonomous bus services can be classified into five main categories: Cluster 1 (red) is primarily about users' attitudes, such as perception, safety, attitudes, concerns, and acceptance toward the service. Cluster 2 (blue) primarily focuses on optimization models, algorithms, and network designs for service systems. Cluster 3 (purple) pertains to the demand for the service and its effect on other services, such as taxis, Ubers, and ride hailing. Cluster 4 (green) primarily pertains to the bus fleet, travel time, and simulation model used in the system optimization analysis. Cluster 5 (yellow) focuses on the behavior of services, mobility, and the effect of land use. In the subsequent chapters, these aspects are systematically reviewed separately.
-
Owing to the numerous advantages of autonomous bus services, a pilot demonstration application was first conducted in Spain as a subproject of the CityMobil project in the EU's 6th Framework Program[16]. Subsequently, autonomous bus services were widely applied in more than a dozen countries and regions in the CityMobil2 project of the EU's 7th Framework Program. Alessandrini et al.[6] conducted a generic SP questionnaire in 12 European cities, where the program was implemented to discover the acceptance of travelers toward conventional and autonomous bus services. The findings indicated that if autonomous bus services cannot provide advantages in terms of trip time and pricing benefits compared with conventional bus services, then travelers will not perceive that they are valuable. From a safety perspective, travelers in these cities tend to be suspicious of automated transit services under mixed-traffic conditions. Herrenkind et al.[19] developed a model to analyze the acceptance of 268 German passengers toward autonomous bus services to demonstrate the advantages of the services. They discovered that a combination of personal factors, social effects, and system characteristics determined the passengers' acceptance. Based on an investigation regarding the usage of autonomous bus services in Greece, Papadima et al.[7] suggested that most residents are satisfied with the service and that price, not safety, is an essential factor in deciding whether to use the service. However, based on the findings of a survey conducted by Roche-Cerasi[14] among 1419 members of the Norwegian Automobile Federation, most individuals were concerned about the safety of autonomous bus services in mixed traffic flows and did not believe that the services offered significant advantages. Unfamiliarity with autonomous bus technology may be a primary factor affecting its acceptance by the general population. According to a field survey of 942 tourists in Germany before and after traveled on an autonomous minibus, Bernhard et al.[9] discovered that most of them were satisfied with the service and that the first-riding experience significantly affected their acceptance of the service.
Based on quantitative travel cost data in Japan, Abe[4] discussed the effect of autonomous bus services on metropolitan transportation systems and demonstrated their potential benefits in reducing passenger travel costs and improving transit accessibility. Sun et al.[12] conducted a survey regarding autonomous bus services in Singapore, which revealed increased service efficiency and decreased total cost of ownership. Moreover, passengers' choice of bus service is significantly determined by travel time and cost, with the latter imposing greater effect. Based on actual travel demand data for Fuyang City in Zhejiang Province, China, Zhai et al.[20] developed an agent-based simulation model to evaluate the effectiveness of autonomous bus services. The results showed that the service required less road space, demonstrated more effective use of bus vehicles, adapted to changes in bus travel demand, and was financially viable. A performance assessment of fixed-route and door-to-door feeder bus services, conducted by Badia & Jenelius[21], revealed that autonomous buses significantly affected service strategies under various technological development scenarios. When autonomous bus technology is sufficiently mature, it can significantly expand the applicability of door-to-door transit services. Using business data before, during, and after the operation of an autonomous bus service in Las Vegas, Kim et al.[11] discovered that employment, education level, and residential employment opportunities significantly affected the preference toward the service and that the service can increase business activity in older, declining downtown areas.
Salonen[13] performed a study on 197 autonomous bus passengers in the Netherlands and showed that the driving safety of autonomous buses was higher than that of regular buses, although passengers felt much less safe inside the vehicles. An analysis of 891 respondents in Philadelphia, USA by Dong et al.[22] revealed that when an employee was on board to oversee the operation or aid, most respondents were willing to ride self-driving buses; otherwise, only 13% indicated willingness. Kassens-Noor et al.[15] surveyed people in Michigan (USA) and discovered that autonomous bus services may attract more travelers. However, approximately one-half of the respondents mentioned they were unsure of reusing the service because they were concerned about safety and did not trust driverless technology. Based on a survey of 1062 respondents in Spain conducted by Rosell & Allen[8], passengers with higher socioeconomic status were more satisfied with autonomous bus services and more willing to reuse them, whereas women were less inclined to ride autonomous buses without employees.
Weschke et al.[23] conducted a Wizard-of-Oz experiment to determine the likelihood of passengers accepting autonomous bus services. They discovered that the services improved feeder trips, whereas travelers wished to be the only ones on the bus and would rather seek services in real time than reserve ahead of travel. Mouratidis & Cobeña Serrano[24] utilized a mixed methodology comprising surveys and interviews to analyze the feedback of passengers before and after they traveled on an autonomous bus and whether they intended to use the service in the future. The findings indicated that most passengers were satisfied with the service and preferred more frequent departures, more rapid vehicle speeds, less abrupt braking, and the presence of employees in the vehicle. An analysis of survey data from Lincoln City in the United States by Piatkowski[25] revealed that autonomous bus services complemented the current bus service system and that the riders' willingness to use the services depended on their age, whether they worked downtown, and their opinion regarding the service. Guo et al.[26] developed a mixed-logit model based on stated-choice experiments in Sweden to determine how passengers select between autonomous and conventional bus services. The findings indicated no significant difference in the selection behavior of these two services under normal conditions; however, the passengers were unwilling to select the autonomous bus service owing to its quality and dependability under unsatisfactory weather conditions. By performing structural equation modeling, Yan et al.[10] analyzed 576 Chinese passengers with experience riding autonomous buses. Empirical research indicates that the presence of a driver, in-vehicle safety, service quality, and good attitude toward the bus positively affect service reuse, whereas road safety does not impose any direct effect. Utilizing survey data from three different time points, Zhao et al.[27] investigated the dynamic characteristics of passenger acceptance toward an automated bus service over time. They discovered that service comfort, frequency, and travel time were crucial in determining whether passengers would continue to use the service. Guo et al.[28] performed structural equation modeling to investigate the acceptance of Stockholm residents toward autonomous bus services on public roadways in a mixed-traffic environment. The findings indicated that the adoption of this service was primarily determined by travel demands and user requirements instead of individual behavioral intentions. The literature related to users' attitudes is summarized in Table 3.
Table 3. Summary of literature related to users' attitudes.
Category Representative studies Data source Analysis approach Acceptance Alessandrini et al. (2016)[6] Questionnaire interview Copula-based logit Herrenkind et al.(2019)[19] Online questionnaire Structural equation model Roche-Cerasi (2019)[14] Questionnaire interview Descriptive statistics Papadima et al. (2020)[7] Online questionnaire Conjoint analysis Bernhard et al. (2020)[9] Questionnaire interview Descriptive statistics Performance Abe (2019)[4] National trip survey in Japan Descriptive statistics Sun et al. (2020)[12] Questionnaire interview Mixed logit Zhai et al. (2020)[20] Travel demand from Zhejiang Agent-based simulation Badia and Jenelius (2021)[21] Virtual operation data Continuum approximation Kim et al. (2022)[11] 3-year business license data Generalized linear mixed-effect regression Perception Salonen (2018)[13] Questionnaire interview Descriptive statistics Dong et al. (2019)[22] Online questionnaire Mixed logit Kassens-Noor et al. (2020)[15] Questionnaire interview Logistic regression Rosell and Allen (2020)[8] Questionnaire interview Structural equation model Preference Weschke et al. (2021)[23] Questionnaire interview Mixed logit model Mouratidis and Serrano (2021)[24] Questionnaire interview Descriptive statistics Piatkowski (2021)[25] Questionnaire interview Ordered logistic regression Guo et al. (2021)[26] Questionnaire interview Mixed logit model Yan et al. (2022)[10] Questionnaire interview Structural equation model Zhao et al. (2022)[27] Questionnaire interview Structural equation model Guo et al. (2022)[28] Questionnaire interview Structural equation model Review of literature related to operations
-
Considering that cities are increasingly deploying autonomous buses in their existing bus service networks, scholars have developed numerous optimization models and solution algorithms to assist authorities in devising optimal operational strategies. Cao et al.[29] developed a multi-objective model while considering schedule deviations, travel time, and energy consumption, which was solved using a multi-criteria decision analysis method based on spatiotemporal graphics techniques. Using optimal hold and speed-changing control strategies, they created autonomous bus schedules with no changes. Considering the requirements of both users and operators, Cao & Avi Ceder[30] devised the skip-stop strategy-based strategy for simultaneously optimizing the service timetable and vehicle schedule of autonomous bus services. Using the defect function graph and considering the constraints on vehicle capacity, they developed a multi-decision choice model and solved it via binary-variable iteration. The results indicated that the skip-stop strategy based on real-time passenger demand reduced the travel time and number of vehicles. Considering autonomous buses as sublines, Gkiotsalitis et al.[31] constructed a mixed-integer planning model to investigate the frequency optimization issue to minimize operation and passenger waiting times. Using the autonomous bus models in the VISSIM simulation platform, Zhang et al.[32] performed case studies to compare the reductions in emissions and energy use for managed and dedicated bus lane strategies. To combine the optimized routes and charging schedules of autonomous buses, Wang et al.[33] developed a constraint programming model with Boolean satisfiability conditions and evaluated the best routes and charger allocation schemes under various constraint scenarios. Zhang et al.[34] constructed a mixed-integer linear program to investigate the lane deployment issues for three travel modes (connected and autonomous buses, connected and autonomous passenger vehicles, and human-driven passenger vehicles) in a transportation network. They insisted that deploying dedicated lanes for autonomous buses would significantly increase the number of passengers and the total benefits to society. Hatzenbühler et al.[35] designed a multi-objective optimization and multi-agent simulation framework to investigate the network design of autonomous bus services under two strategies, i.e., user-centric and operator-centric strategies, to achieve a higher potential travel demand.
Although the application of autonomous buses on a large scale offers considerable benefits, it cannot be accomplished immediately because of financial constraints and uncertainties in the evolution of driverless technology. Based on fluctuations in passenger demand, Dai et al.[36] constructed an integer nonlinear programming model to jointly optimize the capacity and scheduling of systems with autonomous and conventional buses. The bus capacity was varied by assembling or disassembling multiple autonomous minibuses and adjusting the operating time based on forward and backward headways to simultaneously reduce operational and passenger costs. To determine the best fleet size for autonomous buses and distribute vehicles across multiple bus lines, Tian et al.[37] constructed a mixed-integer stochastic programming model reformulated using a quadratic transform with a linear alternating algorithm. Hatzenbühler et al.[38] performed a dynamic transit allocation and developed an operation simulation model to investigate the effect of service frequency and vehicle capacity on the sequential deployment of autonomous buses in a line. Considering passenger acceptability, Tian et al.[39] established a mixed-integer nonlinear model to minimize trip costs to optimize the deployment of autonomous buses on bus lines at various phases. Zhang et al.[40] constructed a generalized operational cost model that accounted for the effects of passenger wait time, ride time, operating cost, and capital cost under various scenarios to determine the parameters that affect the service efficiency provided by autonomous buses.
Recently, owing to advancements in information, communication, and vehicle-road collaboration technologies, several scholars have developed innovative service models using modular autonomous buses. By applying an enhanced deficit function theory, Liu et al.[41] devised a technique to precisely estimate the minimal size of modular autonomous buses required to satisfy the actual demand for public transit trips. Wu et al.[42] proposed a modular autonomous bus system with dynamic transfer strategies to satisfy time-fluctuating traffic demands and constructed a two-stage model to optimize the design of passenger transfer plans and vehicle travel trajectories. Dakic et al.[43] investigated a modular autonomous bus unit combination and scheduling optimization issues based on a three-dimensional macroscopic fundamental diagram to minimize the overall cost of the service system. Based on passenger demand at different periods, Tian et al.[44] constructed a mixed-integer nonlinear program to investigate the optimization of modular autonomous bus stop locations, stop capacities, and vehicle formations, thus demonstrating the benefits of the service in reducing the total cost to operators and passengers. The literature related to operations is summarized in Table 4.
Table 4. Summary of literature related to operations.
Vehicle category Operation Objectives Solution approach Representative studies Autonomous bus Timetable Time-space graphical techniques Cao et al. (2019)[29] Timetable and vehicle schedule GA combined with binary variable iteration Cao & Ceder (2019)[30] Timetable Mixed-integer linear program Gkiotsalitis et al. (2022)[31] Emission and energy Simulation model Zhang et al. (2020)[32] Route and recharge
scheduleConstrained program Wang et al. (2021)[33] Dedicated lanes Mixed-integer nonlinear program Zhang et al. (2022)[34] Network Multi-objective optimization Hatzenbühler et al. (2022)[35] Autonomous and
conventional busSchedule and capacity Mixed-integer nonlinear program Dai et al. (2020)[36] Fleet Mixed-integer stochastic program Tian et al. (2021)[37] Deployment Dynamic assignment and simulation model Hatzenbühler et al. (2020)[38] Deployment Bi-level optimization Tian et al. (2022)[39] Semi- and fully-
autonomous busGeneralized cost Mixed-integer nonlinear program Zhang et al. (2019)[40] Modular autonomous bus Fleet Graphical method Liu et al. (2020)[41] Transfer Two-stage model Wu et al. (2021)[42] Vehicle schedule Three-dimensional macroscopic fundamental diagram Dakic et al. (2021)[43] Station location Mixed-integer nonlinear program Tian et al. (2022)[44] -
The data that support the findings of this study are available in the Web of Science (WoS) repository.
-
About this article
Cite this article
Shen J, Liu Q, Ye Z, Jiang W, Ma C. 2023. Autonomous bus services: current research status and future recommendations. Digital Transportation and Safety 2(3):229−240 doi: 10.48130/DTS-2023-0019
Autonomous bus services: current research status and future recommendations
- Received: 21 April 2023
- Accepted: 18 September 2023
- Published online: 28 September 2023
Abstract: Implementing autonomous bus services in several cities has garnered substantial research attention worldwide. However, the benefits and challenges of this emerging mode remain insufficiently understood. Consequently, VOSviewer was employed for a bibliometric analysis involving 300 publications, investigating the associations among authors, journals, and keywords. Subsequently, we comprehensively reviewed the current state of research on two topics and proposed future recommendations. Results indicate that the first document related to autonomous bus services was published in 2009. Most user attitude -related research data are obtained via questionnaires and analyzed using statistical techniques. Autonomous bus services are expected to benefit passengers regarding travel time, cost, safety, etc., while passenger preferences are inconsistent. However, integrating the service into existing bus systems requires careful consideration of the schedule sequences. Notably, modular autonomous bus services present a new opportunity for the further optimization of bus services. In future studies, standardized data acquisition procedures should be developed to achieve comparable results. Regarding traveler choice behavior, the effect of specific autonomous bus service policies over time and the heterogeneity due to cultural or social contexts across regions should be assessed. To further promote autonomous bus services, based on fluctuating travel demands, the effects of vehicle capacity, speed, and cost of fleet composition should be evaluated comprehensively to optimize the bus network and schedule sequence. Owing to the protracted nature of the transition from conventional to fully autonomous buses, one should prioritize semi-autonomous bus services. Another essential future research direction is to integrate modular autonomous bus assembly or disassembly strategies with different fine-grained operation optimization techniques in various scenarios.
-
Key words:
- Autonomous bus /
- Users' attitude /
- Operation /
- Bibliometric analysis /
- Systematic review /
- VOSviewer