On high-frequency bus routes, telling passengers how crowded upcoming buses are (real-time crowding information, BCI/RTCI) can spread loads across vehicles, improve waiting experience, and dampen bus bunching — but the demonstration underpinning this idea assumed error-free information, idealized operations, and simple at-stop passenger re-choice. The open problem is to establish, and design for, whether and how the system-level benefits of BCI survive realistic conditions: crowding information derived from noisy automatic passenger counts and imperfect load forecasts, passengers with heterogeneous and pre-trip behavioral responses, deliberate dissemination choices, and coordination with holding control beyond simple self-equalizing strategies. Later literature has resolved components (behavioral willingness to wait, occupancy prediction with uncertainty, RTCI–bunching interaction in simulation), but no published work closes the integrated question on buses; a 2024 review of bus bunching confirms the user-centric information line and realistic evaluation of control methods as open gaps.

Background

On a high-frequency bus route (one where buses come every few minutes, so passengers arrive at stops without consulting a timetable), the dominant service-quality failure is bus bunching: a feedback loop in which a slightly late bus picks up more passengers, dwells longer at stops, falls further behind, while the following bus catches up, until buses arrive in pairs. A common remedy is holding — deliberately delaying a bus at a control point so that headways (gaps between successive buses) re-even out; a 'self-equalizing, two-way-looking' holding rule looks at the headways to both the preceding and following buses and holds to equalize them. A complementary, passenger-side remedy is real-time crowding information (RTCI, called BCI for buses): displays or apps telling waiting passengers the expected crowding of the next and following departures, so that some passengers board a less-crowded later bus, voluntarily smoothing loads and headways. Automatic passenger counters (APC) on doors supply boarding/alighting counts from which bus passenger load (BPL) along the route is estimated or predicted; measurement and prediction errors mean the communicated crowding level can deviate from what the passenger actually meets. A stated-preference (SP) survey is a survey technique in which respondents choose between hypothetical travel alternatives, used to estimate behavioral models of, e.g., willingness to wait for a less-crowded vehicle; denial of boarding denotes passengers left behind by a full vehicle. Prior work established the building blocks: Wang, Chen, Chen, Cheng and Lei (2018, doi:10.1177/0361198118798722) simulated how real-time arrival-plus-load information can turn passengers from contributors of headway irregularity into controllers of it; Zhang, Jenelius and Kottenhoff (2016, doi:10.1007/s12469-016-0150-y) piloted crowding information in the Stockholm metro; Drabicki, Kucharski, Cats and Szarata (2020, doi:10.1080/23249935.2020.1809547) simulated network-wide RTCI effects and noted the results are susceptible to information inaccuracy. The direct source is Wang, Chen, Zheng, Cheng, Wang and Lei (2021, doi:10.1016/j.tra.2021.04.007), which formulated bus motion models incorporating passenger boarding choice under BCI and simulated the policy on a holding-controlled high-frequency route, reporting that BCI reduces operation instability and crowding, coordinates well with holding, and remains robust when passengers have low inclinations to choose the next bus. That simulation rested on idealized assumptions: BCI measured without error, all passengers receiving and using the information at the stop, no overtaking, one direction of operation, and only a simple holding rule. The companion prediction study by Wang, Chen, Chen, Hua and Pu (2021, doi:10.1049/itr2.12018) developed a two-stage BPL prediction method (adaptive Kalman filter for short-term stop flows, then support vector regression for loads) from APC data on Suzhou bus line 1 and laid the foundation for accurate and reliable predictive RTCI, leaving bus crowding-level prediction and the release strategy for disseminating RTCI to customers in a reliable and useful way as future work. After 2021, the behavioral side advanced: Drabicki, Cats and Kucharski (2021, doi:10.1109/mt-its49943.2021.9529310) and Drabicki, Kucharski and Cats (2022, doi:10.1007/s11116-022-10270-3) embedded an SP-estimated boarding choice model in a simulation of a Warsaw corridor, finding RTCI induces 30–70% probabilities of skipping an overcrowded bus and substantially reduces headway/load variance and denial-of-boarding; Drabicki, Cats, Kucharski, Fonzone and Szarata (2023, doi:10.1016/j.rtbm.2023.100963) estimated willingness to wait under RTCI; Peftitsi, Jenelius and Cats (2022, doi:10.1016/j.tra.2022.10.011) modeled RTCI-driven passenger distribution in rail, and Leprévost, Portron, Krishnakaran, Grison and Morgagni (2026, doi:10.1016/j.tra.2026.105081) studied platform-placement responses in mass transit. On the information-supply side, Jenelius (2020, doi:10.1109/tits.2019.2914729) quantified data-driven metro crowding prediction error, Noursalehi, Koutsopoulos and Zhao (2021, doi:10.1016/j.trc.2021.103139) built a predictive crowding-information platform for rail, Wood, Yu and Gayah (2023, doi:10.1016/j.ijtst.2022.03.005) evaluated real-time bus occupancy prediction frameworks, and Chen, Cheng, Schmidt and Sun (2025, doi:10.1016/j.trb.2024.103147) forecast bus travel time and occupancy jointly with uncertainty. On the control and information-presentation side, Zhou, Tian and Wang (2022, doi:10.1016/j.tranpol.2022.04.022) explored bunching control utilizing real-time information, Zhang-Kennedy et al. (2023, doi:10.1145/3544548.3581241) studied how passengers perceive crowding visualizations, and the review by Rezazada, Nassir, Tanin and Ceder (2024, doi:10.1080/01441647.2024.2313969) reports that user-centric approaches (steering passengers toward less-crowded vehicles) show some potential while realistic evaluations and real-world applications of control methods remain gaps. What no later work does is connect these strands: every system-level BCI/RTCI evaluation for buses found assumes the displayed crowding is essentially correct, models only at-stop departure re-choice, and treats holding and information as separate.

Problem Statement

For high-frequency bus service, determine whether and how the system-level benefits of real-time bus crowding information (BCI) survive realistic operating conditions, and how BCI provision should be designed accordingly — the question left open by Wang et al. (2021, doi:10.1016/j.tra.2021.04.007), whose simulation assumed error-free BCI, uniformly behaving at-stop passengers, a single direction, no overtaking, and only a simple self-equalizing holding rule. Concretely, realistic conditions comprise four linked directions that remain open individually and jointly; the open problem is the integrated question, so a resolution must address all four directions together and results on any proper subset do not resolve it: (1) generating BCI from an explicitly modeled noisy measurement-and-prediction process (e.g., automatic passenger counting errors and load-forecast uncertainty) and quantifying how the discrepancy between communicated and actual crowding modifies or destroys the policy's benefits relative to the perfect-information ideal; (2) representing passenger behavior beyond uniform at-stop re-choice, including passengers who adjust arrival times using pre-trip information, differing tolerances of waiting time between long-trip and short-trip riders, and passengers who would wait for a later bus even without information; (3) evaluating dissemination design — which passengers receive information, when, and through which channels — in terms of system-level outcomes rather than only individual usefulness; and (4) coordinating BCI with holding control beyond the simple two-way-looking self-equalizing rule, such as optimization-based holding or adaptive control points, including whether joint information-plus-control designs outperform either alone. System-level performance is to be judged on the metrics used in the source line of work — passenger waiting time and its distribution, headway and load regularity, on-board crowding including denial-of-boarding incidence — evaluated against the idealized baseline.

References

  1. Wang, Pengfei; Chen, Xuewu; Zheng, Yue; Cheng, Long; Wang, Yinhai; Lei, Da (2021), 'Providing real-time bus crowding information for passengers: A novel policy to promote high-frequency transit performance', Transportation Research Part A: Policy and Practice 148:316-329, doi:10.1016/j.tra.2021.04.007, https://doi.org/10.1016/j.tra.2021.04.007
  2. Wang, Pengfei; Chen, Xuewu; Chen, Wendong; Cheng, Long; Lei, Da (2018), 'Provision of Bus Real-Time Information: Turning Passengers from Being Contributors of Headway Irregularity to Controllers', Transportation Research Record: Journal of the Transportation Research Board 2672:143-151, doi:10.1177/0361198118798722, https://doi.org/10.1177/0361198118798722
  3. Zhang, Yizhou; Jenelius, Erik; Kottenhoff, Karl (2016), 'Impact of real-time crowding information: a Stockholm metro pilot study', Public Transport 9:483-499, doi:10.1007/s12469-016-0150-y, https://doi.org/10.1007/s12469-016-0150-y
  4. Drabicki, Arkadiusz; Kucharski, Rafal; Cats, Oded; Szarata, Andrzej (2020), 'Modelling the effects of real-time crowding information in urban public transport systems', Transportmetrica A: Transport Science 17:675-713, doi:10.1080/23249935.2020.1809547, https://doi.org/10.1080/23249935.2020.1809547
  5. Drabicki, Arkadiusz; Cats, Oded; Kucharski, Rafal (2021), 'The potential of real-time crowding information in reducing bus bunching under different network saturation levels', 2021 7th International Conference on Models and Technologies for Intelligent Transportation Systems (MT-ITS), 1-6, doi:10.1109/mt-its49943.2021.9529310, https://doi.org/10.1109/mt-its49943.2021.9529310
  6. Drabicki, Arkadiusz; Kucharski, Rafal; Cats, Oded (2022), 'Mitigating bus bunching with real-time crowding information', Transportation 50:1003-1030, doi:10.1007/s11116-022-10270-3, https://doi.org/10.1007/s11116-022-10270-3
  7. Drabicki, Arkadiusz; Cats, Oded; Kucharski, Rafal; Fonzone, Achille; Szarata, Andrzej (2023), 'Should I stay or should I board? Willingness to wait with real-time crowding information in urban public transport', Research in Transportation Business & Management 47:100963, doi:10.1016/j.rtbm.2023.100963, https://doi.org/10.1016/j.rtbm.2023.100963
  8. Peftitsi, Soumela; Jenelius, Erik; Cats, Oded (2022), 'Modeling the effect of real-time crowding information (RTCI) on passenger distribution in trains', Transportation Research Part A: Policy and Practice 166:354-368, doi:10.1016/j.tra.2022.10.011, https://doi.org/10.1016/j.tra.2022.10.011
  9. Jenelius, Erik (2020), 'Data-Driven Metro Train Crowding Prediction Based on Real-Time Load Data', IEEE Transactions on Intelligent Transportation Systems 21:2254-2265, doi:10.1109/tits.2019.2914729, https://doi.org/10.1109/tits.2019.2914729
  10. Wood, Jonathan; Yu, Zhengyao; Gayah, Vikash V. (2023), 'Development and evaluation of frameworks for real-time bus passenger occupancy prediction', International Journal of Transportation Science and Technology 12:399-413, doi:10.1016/j.ijtst.2022.03.005, https://doi.org/10.1016/j.ijtst.2022.03.005
  11. Chen, Xiaoxu; Cheng, Zhanhong; Schmidt, Alexandra M.; Sun, Lijun (2025), 'Conditional forecasting of bus travel time and passenger occupancy with Bayesian Markov regime-switching vector autoregression', Transportation Research Part B: Methodological 192:103147, doi:10.1016/j.trb.2024.103147, https://doi.org/10.1016/j.trb.2024.103147
  12. Noursalehi, Peyman; Koutsopoulos, Haris N.; Zhao, Jinhua (2021), 'Predictive decision support platform and its application in crowding prediction and passenger information generation', Transportation Research Part C: Emerging Technologies 129:103139, doi:10.1016/j.trc.2021.103139, https://doi.org/10.1016/j.trc.2021.103139
  13. Zhou, Chang; Tian, Qiong; Wang, David Z.W. (2022), 'A novel control strategy in mitigating bus bunching: Utilizing real-time information', Transport Policy 123:1-13, doi:10.1016/j.tranpol.2022.04.022, https://doi.org/10.1016/j.tranpol.2022.04.022
  14. Rezazada, Mustafa; Nassir, Neema; Tanin, Egemen; Ceder, Avishai (Avi) (2024), 'Bus bunching: a comprehensive review from demand, supply, and decision-making perspectives', Transport Reviews 44:766-790, doi:10.1080/01441647.2024.2313969, https://doi.org/10.1080/01441647.2024.2313969
  15. Leprévost, Florian; Portron, Arthur; Krishnakaran, Kirusha; Grison, Élise; Morgagni, Simone (2026), 'Impact of real-time crowding information on mass-transit passengers' platform placement choices', Transportation Research Part A: Policy and Practice 211:105081, doi:10.1016/j.tra.2026.105081, https://doi.org/10.1016/j.tra.2026.105081
  16. Zhang-Kennedy, Leah; Aziz, Saira; Oluwadare, Oluwafunminitemi (Temi); Pan, Lyndon; Wu, Zeyu; Lamorea, Sydney E.C.; Li, Soda; Sun, Michael; Mäkelä, Ville (2023), 'Passenger Perceptions, Information Preferences, and Usability of Crowding Visualizations on Public Displays in Transit Stations and Vehicles', Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 1-15, doi:10.1145/3544548.3581241, https://doi.org/10.1145/3544548.3581241
  17. Wang, Pengfei; Chen, Xuewu; Chen, Jingxu; Hua, Mingzhuang; Pu, Ziyuan (2021), 'A two-stage method for bus passenger load prediction using automatic passenger counting data', IET Intelligent Transport Systems 15:248-260, doi:10.1049/itr2.12018, https://doi.org/10.1049/itr2.12018