Pub Date : 2026-01-02Epub Date: 2024-08-15DOI: 10.1080/15472450.2024.2392722
İlyas Cihan Aksoy , Mehmet Metin Mutlu
The Transit Network Frequency Setting Problem (TNFSP), an NP-Hard combinatorial optimization problem, has been frequently addressed in previous investigations, most of which employ metaheuristics. However, previous studies have not investigated to determine the best metaheuristic and the most effective parameter values for the TNFSP. This study aims to fill this research gap by comprehensively comparing five well-known metaheuristics, namely, Artificial Bee Colony, Differential Evolution, Firefly Algorithm, Genetic Algorithm, and Particle Swarm Optimization, on the TNFSP. The best parameter configurations for employed metaheuristics are determined using the brute-force method. The comparative results not only highlight the significance of metaheuristic selection in addressing the TNFSP but also reveal the importance of choosing the most suitable parameter values for these metaheuristics. Furthermore, the metaheuristics are performed on the transit networks obtained from previously published studies for comparison purposes. The results demonstrate that the metaheuristics achieve better frequency sets than the original frequency sets of the used transit networks.
{"title":"Comparing the performance of metaheuristics on the Transit Network Frequency Setting Problem","authors":"İlyas Cihan Aksoy , Mehmet Metin Mutlu","doi":"10.1080/15472450.2024.2392722","DOIUrl":"10.1080/15472450.2024.2392722","url":null,"abstract":"<div><div>The Transit Network Frequency Setting Problem (TNFSP), an NP-Hard combinatorial optimization problem, has been frequently addressed in previous investigations, most of which employ metaheuristics. However, previous studies have not investigated to determine the best metaheuristic and the most effective parameter values for the TNFSP. This study aims to fill this research gap by comprehensively comparing five well-known metaheuristics, namely, Artificial Bee Colony, Differential Evolution, Firefly Algorithm, Genetic Algorithm, and Particle Swarm Optimization, on the TNFSP. The best parameter configurations for employed metaheuristics are determined using the brute-force method. The comparative results not only highlight the significance of metaheuristic selection in addressing the TNFSP but also reveal the importance of choosing the most suitable parameter values for these metaheuristics. Furthermore, the metaheuristics are performed on the transit networks obtained from previously published studies for comparison purposes. The results demonstrate that the metaheuristics achieve better frequency sets than the original frequency sets of the used transit networks.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 1-20"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142247329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-09-24DOI: 10.1080/15472450.2024.2408024
Sania E. Seilabi , Mohammadhosein Pourgholamali , Mohammad Miralinaghi , Gonçalo Homem de Almeida Correia , Xiaozheng (Sean) He , Samuel Labi
The introduction of connected and autonomous vehicles (CAVs) provides a significant opportunity to address the persistently increasing problem of urban traffic congestion. By virtue of their connectivity and automation features, CAVs can reduce vehicle headways, thereby increasing road capacity and enhancing throughput. It has been hypothesized that CAV-infrastructure design policies can influence traveler behavior in ways that could reduce congestion. This research focuses on the potential of using CAV-dedicated lanes (CAVL) to alleviate traffic congestion in a bottleneck corridor that serves both human-driven vehicles (HDVs) and CAVs. We delve into investigating the impacts of CAVLs on the departure time and lane choices of morning commuters. The study first expresses traffic equilibrium conditions as a linear program with complementarity constraints. Then, a system-optimal commute congestion management design is formulated to minimize the overall system cost, which consists of queuing delays and early and late arrival costs. The results of the computational experiments suggest that: (i) the CAV technological advancements can significantly reduce traffic congestion under CAVL deployment with an almost similar effect as a tolling policy; and (ii) the lower value of time for CAV commuters leads them to depart closer to their desired arrival time without a tolling policy, which could significantly increase the bottleneck traffic congestion that commuters experience, particularly HDVs.
{"title":"Optimizing dedicated lanes and tolling schemes for connected and autonomous vehicles to address bottleneck congestion considering morning commuter departure choices","authors":"Sania E. Seilabi , Mohammadhosein Pourgholamali , Mohammad Miralinaghi , Gonçalo Homem de Almeida Correia , Xiaozheng (Sean) He , Samuel Labi","doi":"10.1080/15472450.2024.2408024","DOIUrl":"10.1080/15472450.2024.2408024","url":null,"abstract":"<div><div>The introduction of connected and autonomous vehicles (CAVs) provides a significant opportunity to address the persistently increasing problem of urban traffic congestion. By virtue of their connectivity and automation features, CAVs can reduce vehicle headways, thereby increasing road capacity and enhancing throughput. It has been hypothesized that CAV-infrastructure design policies can influence traveler behavior in ways that could reduce congestion. This research focuses on the potential of using CAV-dedicated lanes (CAVL) to alleviate traffic congestion in a bottleneck corridor that serves both human-driven vehicles (HDVs) and CAVs. We delve into investigating the impacts of CAVLs on the departure time and lane choices of morning commuters. The study first expresses traffic equilibrium conditions as a linear program with complementarity constraints. Then, a system-optimal commute congestion management design is formulated to minimize the overall system cost, which consists of queuing delays and early and late arrival costs. The results of the computational experiments suggest that: (i) the CAV technological advancements can significantly reduce traffic congestion under CAVL deployment with an almost similar effect as a tolling policy; and (ii) the lower value of time for CAV commuters leads them to depart closer to their desired arrival time without a tolling policy, which could significantly increase the bottleneck traffic congestion that commuters experience, particularly HDVs.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 37-54"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147855399","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-10-17DOI: 10.1080/15472450.2024.2418902
Anne S. Patricio , Gonçalo Gonçalves Duarte Santos , António Pais Antunes
Driverless vehicles have been attracting the attention of many researchers due to their potential positive impacts on current travel patterns and transport sustainability. In a regional context, driverless demand-responsive transit (DRT) systems are expected to benefit both passengers and operators, currently affected by low service levels and cost inefficiencies. Low-cost DRT systems may enable lower fares and consequently attract more users. The use of a hub-and-spoke (H&S) DRT solution can be viewed as a cheaper alternative to point-to-point (without the use of hubs) services, though there is an absence of literature contributions comparing DRT systems with these two designs. In this paper, we perform a preliminary assessment on how a H&S design can impact the performance of a driverless DRT regional system, considering the perspectives of the system’s operator and of its passengers. For this purpose, we rely on a methodological approach through which a H&S design based on a hub location model is compared to a point-to-point design. The approach is applied to analyze system operations in a region in central Portugal. The results indicate that a H&S design can increase the profits involved in the operation of the system, especially when compared to a point-to-point system without ride-pooling. However, profit increases come with a cost for passengers in terms of travel times and distances.
{"title":"A preliminary assessment on the design of driverless demand-responsive transit regional systems: point-to-point vs. hub-and-spoke","authors":"Anne S. Patricio , Gonçalo Gonçalves Duarte Santos , António Pais Antunes","doi":"10.1080/15472450.2024.2418902","DOIUrl":"10.1080/15472450.2024.2418902","url":null,"abstract":"<div><div>Driverless vehicles have been attracting the attention of many researchers due to their potential positive impacts on current travel patterns and transport sustainability. In a regional context, driverless demand-responsive transit (DRT) systems are expected to benefit both passengers and operators, currently affected by low service levels and cost inefficiencies. Low-cost DRT systems may enable lower fares and consequently attract more users. The use of a hub-and-spoke (H&S) DRT solution can be viewed as a cheaper alternative to point-to-point (without the use of hubs) services, though there is an absence of literature contributions comparing DRT systems with these two designs. In this paper, we perform a preliminary assessment on how a H&S design can impact the performance of a driverless DRT regional system, considering the perspectives of the system’s operator and of its passengers. For this purpose, we rely on a methodological approach through which a H&S design based on a hub location model is compared to a point-to-point design. The approach is applied to analyze system operations in a region in central Portugal. The results indicate that a H&S design can increase the profits involved in the operation of the system, especially when compared to a point-to-point system without ride-pooling. However, profit increases come with a cost for passengers in terms of travel times and distances.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 129-148"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147862868","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-09-09DOI: 10.1080/15472450.2024.2402699
Mohammad Iqbal , Farid Kurniatama , Mohammad Isa Irawan , Imam Mukhlash , Bandung Arry Sanjoyo
We propose a sequence-to-sequence model to capture a partial trajectory that contains a few attractive local regions and forecast its destination. Therefore, the main goal of this work is to predict the destination, given the trip’s trajectory, effectively by learning from its attractive local sub-trajectory only. Conceptually, the proposed model incorporates Bidirectional Long Short-Term Memory (BiLSTM) for finding frequent trip trajectories and a Consecutive Attractive Local Regions (CALR) mechanism for only taking their few attractive local regions, called CALR-BiLSTM. Existing related works focused on observing whole regions of each trip trajectory at once to decide its attractive ones from a Global Attention Mechanism (GAM). However, GAM may hold similar attractive sub-trajectories for different destinations, which leads to worse predictions. To overcome the issue, the proposed model observes the entire trip trajectory part by part within a small window, such as focusing on a few regions that are attractive. In this work, we demonstrated the proposed model on large public datasets. As a result, we can enjoy the proposed model as the winner against the state-of-the-art models. Moreover, we present ablation studies on various recurrent neural networks and attention mechanisms to ensure the proposed model is in the right settings.
{"title":"Consecutive attractive local regions – bidirectional long short-term memory for trip destination prediction","authors":"Mohammad Iqbal , Farid Kurniatama , Mohammad Isa Irawan , Imam Mukhlash , Bandung Arry Sanjoyo","doi":"10.1080/15472450.2024.2402699","DOIUrl":"10.1080/15472450.2024.2402699","url":null,"abstract":"<div><div>We propose a sequence-to-sequence model to capture a partial trajectory that contains a few attractive local regions and forecast its destination. Therefore, the main goal of this work is to predict the destination, given the trip’s trajectory, effectively by learning from its attractive local sub-trajectory only. Conceptually, the proposed model incorporates Bidirectional Long Short-Term Memory (BiLSTM) for finding frequent trip trajectories and a Consecutive Attractive Local Regions (CALR) mechanism for only taking their few attractive local regions, called CALR-BiLSTM. Existing related works focused on observing whole regions of each trip trajectory at once to decide its attractive ones from a Global Attention Mechanism (GAM). However, GAM may hold similar attractive sub-trajectories for different destinations, which leads to worse predictions. To overcome the issue, the proposed model observes the entire trip trajectory part by part within a small window, such as focusing on a few regions that are attractive. In this work, we demonstrated the proposed model on large public datasets. As a result, we can enjoy the proposed model as the winner against the state-of-the-art models. Moreover, we present ablation studies on various recurrent neural networks and attention mechanisms to ensure the proposed model is in the right settings.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 21-36"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147855398","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-10-21DOI: 10.1080/15472450.2024.2418922
Bo Liu , Liguang Li , Hua Huang , Xitong Zhang
The one-time investment and later operation efficiency of the supporting charging station are critical for the length of the investment return cycle. In this article, to meet the charging demand of logistics parks as the precondition, through the minimum total cost (construction, operation, maintenance, and depreciation) as the objective function, the improved genetic algorithm (GA) is used to look for the optimal solution. The operations of the charging station planning of a logistics park in off-season and peak-season are simulated and analyzed. The results show that the model and calculation method proposed in this article can provide effective support for the planning decision of logistics park charging. This study can guide the improvement and management in charging station configuration.
{"title":"Logistics park charging station optimization using improved genetic algorithms","authors":"Bo Liu , Liguang Li , Hua Huang , Xitong Zhang","doi":"10.1080/15472450.2024.2418922","DOIUrl":"10.1080/15472450.2024.2418922","url":null,"abstract":"<div><div>The one-time investment and later operation efficiency of the supporting charging station are critical for the length of the investment return cycle. In this article, to meet the charging demand of logistics parks as the precondition, through the minimum total cost (construction, operation, maintenance, and depreciation) as the objective function, the improved genetic algorithm (GA) is used to look for the optimal solution. The operations of the charging station planning of a logistics park in off-season and peak-season are simulated and analyzed. The results show that the model and calculation method proposed in this article can provide effective support for the planning decision of logistics park charging. This study can guide the improvement and management in charging station configuration.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 149-156"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147855404","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-11-09DOI: 10.1080/15472450.2024.2425297
Shian Wang , Michael W. Levin , Raphael Stern
Emerging automated vehicle (AV) technologies open a door for cyberattacks, where a select number of AVs are compromised to drive in an adversarial manner, degrading the performance of transportation systems. Hence, designing stable controls for AVs that can mitigate the impact of attacks on traffic flow is important and necessary as AVs gradually become a reality. In this study, we first present a general framework for describing mixed-autonomy traffic involving AVs and human-driven vehicles (HVs) based on car-following dynamics. Under this framework, a class of malicious attacks on AV control commands (vehicle acceleration) is mathematically characterized. To deal with the lack of knowledge of the attacks, we model attacks as an unstructured process, allowing for the inclusion of both deterministic and stochastic attack behaviors, bounded by a known bound (to remain stealthy) without being subject to any specific statistical distribution. Moreover, we analytically derive a set of sufficient conditions for string stable control design of individual AVs which can help mitigate the disturbances to traffic flow caused by unknown attacks on AVs. Furthermore, for any given market penetration rate of AVs, a set of conditions is also derived for selecting appropriate feedback gains of AVs to ensure string stability of heterogeneous traffic, reducing the undesired impact of attacks on traffic flow. These sufficient conditions provide important criteria for string stable control design of AVs without requiring much knowledge of the attacks. A series of numerical results is presented to show effectiveness of the proposed approach on mitigating attack disruption.
{"title":"String stable control design for automated vehicles under cyberattacks","authors":"Shian Wang , Michael W. Levin , Raphael Stern","doi":"10.1080/15472450.2024.2425297","DOIUrl":"10.1080/15472450.2024.2425297","url":null,"abstract":"<div><div>Emerging automated vehicle (AV) technologies open a door for cyberattacks, where a select number of AVs are compromised to drive in an adversarial manner, degrading the performance of transportation systems. Hence, designing stable controls for AVs that can mitigate the impact of attacks on traffic flow is important and necessary as AVs gradually become a reality. In this study, we first present a general framework for describing mixed-autonomy traffic involving AVs and human-driven vehicles (HVs) based on car-following dynamics. Under this framework, a class of malicious attacks on AV control commands (vehicle acceleration) is mathematically characterized. To deal with the lack of knowledge of the attacks, we model attacks as an unstructured process, allowing for the inclusion of both deterministic and stochastic attack behaviors, bounded by a known bound (to remain stealthy) without being subject to any specific statistical distribution. Moreover, we analytically derive a set of sufficient conditions for string stable control design of individual AVs which can help mitigate the disturbances to traffic flow caused by unknown attacks on AVs. Furthermore, for any given market penetration rate of AVs, a set of conditions is also derived for selecting appropriate feedback gains of AVs to ensure string stability of heterogeneous traffic, reducing the undesired impact of attacks on traffic flow. These sufficient conditions provide important criteria for string stable control design of AVs without requiring much knowledge of the attacks. A series of numerical results is presented to show effectiveness of the proposed approach on mitigating attack disruption.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 177-189"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147855406","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-10-12DOI: 10.1080/15472450.2024.2416165
Taniya Sultana , Hany M. Hassan , Brian Wolshon
Vehicle-to-infrastructure (V2I) communication is a key component of the emerging Connected and Automated Vehicles (CAVs) technology. Although previous studies have investigated driver behaviors toward V2I messages, there remains a gap in understanding the underlying factors behind driver perspectives and preferences regarding this communication technology. This paper aims to examine drivers’ preferences, encompassing both their willingness to purchase V2I-equipped vehicles and their preferences for receiving V2I messages, as well as their perceptions of these messages. Additionally, the contributing factors behind these perceptions and preferences were also determined. Data from an online national survey of 1563 drivers from across the United States was analyzed using descriptive statistics and structural equation modeling. Descriptive statistics showed that over 83% of the respondents perceived V2I warning messages as ‘important or extremely important’ to their driving. The structural equation modeling revealed that gender, age, driving experience, commute time, vehicle ownership, education, and prior crash involvement significantly influence drivers’ preferences for V2I communication. The results suggested that individuals who are younger, female, novice, highly educated, long-time commuters with prior crash involvement and lower vehicle ownership are more likely to buy future V2I-equipped vehicles. These same groups also showed increased preference for receiving V2I messages at multiple locations and through multiple methods. Moreover, drivers with favorable views of existing in-vehicle safety-related features (e.g., blind-spot assistance.) demonstrated a higher inclination to own future V2I-equipped vehicles. By providing valuable insights on drivers’ preferences and perceptions, this study contributes to improved planning for the integration of V2I technology in transportation systems.
{"title":"Toward the integration of vehicle-to-infrastructure (V2I) communication in transportation system: an investigation of drivers’ perceptions and preferences for V2I messages","authors":"Taniya Sultana , Hany M. Hassan , Brian Wolshon","doi":"10.1080/15472450.2024.2416165","DOIUrl":"10.1080/15472450.2024.2416165","url":null,"abstract":"<div><div>Vehicle-to-infrastructure (V2I) communication is a key component of the emerging Connected and Automated Vehicles (CAVs) technology. Although previous studies have investigated driver behaviors toward V2I messages, there remains a gap in understanding the underlying factors behind driver perspectives and preferences regarding this communication technology. This paper aims to examine drivers’ preferences, encompassing both their willingness to purchase V2I-equipped vehicles and their preferences for receiving V2I messages, as well as their perceptions of these messages. Additionally, the contributing factors behind these perceptions and preferences were also determined. Data from an online national survey of 1563 drivers from across the United States was analyzed using descriptive statistics and structural equation modeling. Descriptive statistics showed that over 83% of the respondents perceived V2I warning messages as ‘important or extremely important’ to their driving. The structural equation modeling revealed that gender, age, driving experience, commute time, vehicle ownership, education, and prior crash involvement significantly influence drivers’ preferences for V2I communication. The results suggested that individuals who are younger, female, novice, highly educated, long-time commuters with prior crash involvement and lower vehicle ownership are more likely to buy future V2I-equipped vehicles. These same groups also showed increased preference for receiving V2I messages at multiple locations and through multiple methods. Moreover, drivers with favorable views of existing in-vehicle safety-related features (e.g., blind-spot assistance.) demonstrated a higher inclination to own future V2I-equipped vehicles. By providing valuable insights on drivers’ preferences and perceptions, this study contributes to improved planning for the integration of V2I technology in transportation systems.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 69-91"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147855401","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-10-21DOI: 10.1080/15472450.2024.2417175
Diego Da Silva , Amer Shalaby
Accurately monitoring passenger demand fluctuations is crucial for streamlined operations of subway systems and informed decision-making. This study presents a detailed Time Series Analysis of the Toronto subway system using Wi-Fi data connection from devices as a predictor of passenger volume. Various time series models were tested for short-term forecasting, including Linear Regression, Exponential Smoothing, ARIMA, Random Forest, N-BEATS, and T-GCN. An end-to-end modeling implementation process was carried out, and the performance of each model was evaluated. The primary objective was to assess the effectiveness of short-term prediction models for univariate time series at the system level and discuss deployment challenges. While conventional time series models are fast to implement and interpretable, they require a more in-depth data exploration phase for validation, making scaling at the system level difficult. Additionally, maintenance is more challenging with conventional models, and their exploratory analysis phases need to be repeated when the models degrade over time. Prediction difficulty varied across each subway station, indicating the need for a more thorough calibration or hybrid approach, especially for transfer stations. Despite the different uses and qualities of each model in our scenario, Random Forest and Exponential Smoothing emerged as the best performers and could be a satisfactory option for robust demand forecasting at the system level.
{"title":"Forecasting short-term subway passenger flow using Wi-Fi data: comparative analysis of advanced time-series methods","authors":"Diego Da Silva , Amer Shalaby","doi":"10.1080/15472450.2024.2417175","DOIUrl":"10.1080/15472450.2024.2417175","url":null,"abstract":"<div><div>Accurately monitoring passenger demand fluctuations is crucial for streamlined operations of subway systems and informed decision-making. This study presents a detailed Time Series Analysis of the Toronto subway system using Wi-Fi data connection from devices as a predictor of passenger volume. Various time series models were tested for short-term forecasting, including Linear Regression, Exponential Smoothing, ARIMA, Random Forest, N-BEATS, and T-GCN. An end-to-end modeling implementation process was carried out, and the performance of each model was evaluated. The primary objective was to assess the effectiveness of short-term prediction models for univariate time series at the system level and discuss deployment challenges. While conventional time series models are fast to implement and interpretable, they require a more in-depth data exploration phase for validation, making scaling at the system level difficult. Additionally, maintenance is more challenging with conventional models, and their exploratory analysis phases need to be repeated when the models degrade over time. Prediction difficulty varied across each subway station, indicating the need for a more thorough calibration or hybrid approach, especially for transfer stations. Despite the different uses and qualities of each model in our scenario, Random Forest and Exponential Smoothing emerged as the best performers and could be a satisfactory option for robust demand forecasting at the system level.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 108-128"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147855403","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-11-10DOI: 10.1080/15472450.2024.2425293
Victor Jian Ming Low , Hooi Ling Khoo , Wooi Chen Khoo
This study addresses the multifaceted challenge of ensuring the regularity of bus services, minimizing bus bunching, and facilitating synchronized bus connections across routes. An enhanced multi-agent reinforcement learning algorithm, namely the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm, is proposed to implement real-time control strategies for addressing these issues simultaneously. The merit of the modified MADDPG algorithm lies in its ability to continuously learn while adeptly navigating the non-stationary operating nature of bus system networks. A case study of a bus corridor is used to train and test the algorithm. Four robust scenarios, each presenting varying degrees of travel time and dwell time variations, are designed to assess the algorithm’s robustness. Results indicate that the MADDPG algorithm can significantly increase the likelihood of synchronized bus transfers across multiple routes by two or three times while maintaining the service reliability on each route. Moreover, the flexibility of the MADDPG algorithm in training bus policies allows it to effectively adapt to up to 90% variations in bus travel times and demand changes, even amid disruptive events in real-world scenarios.
本研究解决了多方面的挑战，包括确保巴士服务的规律性，减少巴士拥挤，以及促进巴士跨路线的同步连接。提出了一种增强型多智能体强化学习算法，即多智能体深度确定性策略梯度（multi-agent Deep Deterministic Policy Gradient， MADDPG）算法，以实现同时解决这些问题的实时控制策略。改进后的madpg算法的优点在于它具有持续学习的能力，同时能够熟练地驾驭公交系统网络的非平稳运行特性。以公交走廊为例，对该算法进行了训练和测试。设计了四个鲁棒场景，每个场景都呈现不同程度的旅行时间和停留时间变化，以评估算法的鲁棒性。结果表明，madpg算法在保证每条线路的服务可靠性的前提下，可将多路公交同步换乘的可能性显著提高2 ~ 3倍。此外，madpg算法在训练公交策略方面的灵活性使其能够有效地适应高达90%的公交行驶时间和需求变化，即使在现实世界的破坏性事件中也是如此。
{"title":"Robust dynamic real-time control strategies for high-frequency bus service: a multi-agent reinforcement learning framework","authors":"Victor Jian Ming Low , Hooi Ling Khoo , Wooi Chen Khoo","doi":"10.1080/15472450.2024.2425293","DOIUrl":"10.1080/15472450.2024.2425293","url":null,"abstract":"<div><div>This study addresses the multifaceted challenge of ensuring the regularity of bus services, minimizing bus bunching, and facilitating synchronized bus connections across routes. An enhanced multi-agent reinforcement learning algorithm, namely the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm, is proposed to implement real-time control strategies for addressing these issues simultaneously. The merit of the modified MADDPG algorithm lies in its ability to continuously learn while adeptly navigating the non-stationary operating nature of bus system networks. A case study of a bus corridor is used to train and test the algorithm. Four robust scenarios, each presenting varying degrees of travel time and dwell time variations, are designed to assess the algorithm’s robustness. Results indicate that the MADDPG algorithm can significantly increase the likelihood of synchronized bus transfers across multiple routes by two or three times while maintaining the service reliability on each route. Moreover, the flexibility of the MADDPG algorithm in training bus policies allows it to effectively adapt to up to 90% variations in bus travel times and demand changes, even amid disruptive events in real-world scenarios.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 157-176"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147855405","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-02Epub Date: 2024-11-04DOI: 10.1080/15472450.2024.2425304
Jiming Xie , Yan Zhang , Yaqin Qin , Ke Li , Shuai Dong , Siyu Liu , Yulan Xia
Enhancing the synergy between autonomous and human-driven vehicles at the societal level requires understanding drivers’ behaviors and cognitive patterns, as well as conducting human-like driving intention analysis. To achieve this goal, this study designs a novel framework for analyzing human-like driving intention. Firstly, a spectral clustering method is employed to characterize driving styles. Secondly, a misclassification cost matrix is tailored to different driving needs. Finally, inspired by the complex neural networks found in the human brain, we develop a specialized lightweight neural network, termed the Width Human-like Neural Network (WNN), aimed at realizing personalized cognition and facilitating human-like decision-making in driving intention. Experimental studies and validation based on natural driving trajectory data from Kunming, China, demonstrate that the method accurately infers internal implicit driving intention from external explicit and observable driving behaviors, achieving a prediction accuracy of 99.8%. This framework strategically allocates limited computational resources to critical areas for autonomous vehicles and exemplifies best practices for improving neural network performance in driving intention analysis tasks.
{"title":"Wide human-like neural network incorporating driving styles for human-like driving intention analysis","authors":"Jiming Xie , Yan Zhang , Yaqin Qin , Ke Li , Shuai Dong , Siyu Liu , Yulan Xia","doi":"10.1080/15472450.2024.2425304","DOIUrl":"10.1080/15472450.2024.2425304","url":null,"abstract":"<div><div>Enhancing the synergy between autonomous and human-driven vehicles at the societal level requires understanding drivers’ behaviors and cognitive patterns, as well as conducting human-like driving intention analysis. To achieve this goal, this study designs a novel framework for analyzing human-like driving intention. Firstly, a spectral clustering method is employed to characterize driving styles. Secondly, a misclassification cost matrix is tailored to different driving needs. Finally, inspired by the complex neural networks found in the human brain, we develop a specialized lightweight neural network, termed the Width Human-like Neural Network (WNN), aimed at realizing personalized cognition and facilitating human-like decision-making in driving intention. Experimental studies and validation based on natural driving trajectory data from Kunming, China, demonstrate that the method accurately infers internal implicit driving intention from external explicit and observable driving behaviors, achieving a prediction accuracy of 99.8%. This framework strategically allocates limited computational resources to critical areas for autonomous vehicles and exemplifies best practices for improving neural network performance in driving intention analysis tasks.</div></div>","PeriodicalId":54792,"journal":{"name":"Journal of Intelligent Transportation Systems","volume":"30 1","pages":"Pages 190-211"},"PeriodicalIF":2.8,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147855407","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号