Individual Travel Research Articles

Day-to-day traffic control for networks mixed with regular human-piloted and connected autonomous vehicles

The emerging connected autonomous vehicle (CAV) technology offers a significant opportunity to address the congestion problem caused by the selfish routing behavior of human drivers. If the transportation management center (TMC) can control the routing of all vehicles on the network, they could easily reverse this effect. However, during the transition period, CAVs will share roads with human-piloted vehicles (HVs). In the era of mixed autonomy, a more realistic scenario is to route the CAVs to lead the traffic state towards the desired equilibrium. We first consider a repeated routing game in which the TMC can control the routes of CAVs to minimize the system travel time, i.e., CAVs follow the system optimal (SO) routing principle. In this manner, we can influence the travel time traversing each link, which indirectly influences HV users’ route choices. As a consequence, HV users may adjust their routes dynamically through day-to-day (DTD) learning to minimize individual travel cost. On the other hand, the SO routing strategy may induce higher travel time for CAVs than HVs within the same origin–destination (OD) pair. To address such inherent unfairness, road pricing is used to influence route choices of HVs to achieve specific supply regulation targets, e.g., traffic restraint. We present qualitative analysis on the stability of the controlled DTD dynamics and the effect of road pricing regarding the unfairness. To the best of our knowledge, this paper is one of the first to design dynamic pricing schemes for a mixed autonomy system under DTD learning dynamics. We carry out numerical experiments to demonstrate the effectiveness of the control schemes. The cooperative CAVs could reduce the total travel time when the market penetration rate of CAVs reaches the lower minimum control ratio that the network could escape the user equilibrium (UE) condition. When the market penetration rate of CAVs reaches the upper minimum control ratio, the network would achieve SO condition even though there are non-cooperative HVs. On the other hand, the pricing scheme can reduce the unfairness index and total travel time of the network when the market penetration rate of CAVs is low.

Pollution characteristics and source differences of VOCs before and after COVID-19 in Beijing

During the outbreak of the COVID-19, the change in the way of people's living and production provided the opportunity to study the influence of human activity on Volatile organic compounds (VOCs) in the atmosphere. Therefore, this study analyzed VOCs concentration and composition characteristics in urban area of Beijing from 2019 to 2020. The results showed that the concentration of VOCs in Chaoyang district in 2020 was 73.1ppbv, lower than that in 2019 (92.8ppbv), and alkanes (45 % and 47 %) were the most dominant components. The concentrations of isopentane, n-pentane, n-hexane, and OVOCs significantly increased in 2020. According to the results of the PMF model, the contribution of VOCs from vehicle and pharmaceutical-related emissions increased to 45.8 % and 27.1 % in 2020, while coal combustion decreased by 23.7 %. This is likely linked to the strict implementation of the coal conversion policy, as well as the increment in individual travel and pharmaceutical production during the pandemic. The calculation results of OFP and SOAFP indicated that toluene had an increased impact on the formation of O3 and SOA in the Chaoyang district in 2020. Notably, VOCs emitted by vehicles have the highest potential for secondary generation. In addition, VOCs from vehicles and industries pose the greatest health risks, together accounting for 77.4 % and 79.31 % of the total carcinogenic risk in 2019 and 2020. Although industrial emission with the high proportions of halocarbons was controlled to some extent during the pandemic, the carcinogenic risk in 2020 was 3.74 × 10−6, which still exceeded the acceptable level, and more attention and governance efforts should be given to.

A Spatial Network-Based Assessment of Individual Exposure to COVID-19

This study seeks to examine the possible impacts of sociodemographic factors, individual mobility patterns, and daily activities on individual exposure to COVID-19 risk when assessed by different risk measures. Taking Hong Kong as the study area, we first model the risk of COVID-19 using a density-based approach and a network-based approach and reveal the differences in the spatial distributions of COVID-19 transmission risk they obtained. Then, using two-day individual Global Positioning System trajectory data and travel diaries collected from individuals in two communities, we measure individual exposure to COVID-19 transmission risk based on their mobility patterns and reveal the disparities in COVID-19 risk exposure among residents of different demographic groups and neighborhoods when conducting different activities. This study reveals the differences in the spatial patterns of COVID-19 transmission risk when the risk is conceptualized by density and network. It demonstrates the power of a spatial network approach in enhancing our understanding of individual exposure to infectious diseases. This study also advances the existing literature by exploring COVID-19 risk exposure, considering individual daily mobility and addressing the uncertain geographic context problem in analyzing individual COVID-19 risk exposure.

How do I want the city council to spend our budget? Conceiving MaaS from a citizen's perspective … (as well as biking infrastructure and public transport)

While several governmental and research efforts are set upon mobility-as-a-service (MaaS), most of them are driven by individual travel behavior and potential usage. However, considering only individuals’ preferences carries the risk of neglecting societal benefits going beyond individual travel behavior. This study addresses the valuation of different features of MaaS-services from a social desirability perspective as compared to social investments in biking infrastructure and in public transport, and aims at eliciting trade-offs between different features of such projects. This analysis is conducted on the basis of Participatory Value Evaluation (PVE). In PVE-experiments, individuals select their preferred portfolio of government projects given a constrained public budget and societal preferences for (the impacts of) government projects can be determined based on these choices.The results show that the population of Rotterdam exhibits a willingness to allocate public resources to all types of investment projects considered in the analysis. However, the willingness to allocate resources to bike infrastructure projects and public transport seems to be higher than the willingness to dedicate resources to MaaS subsidies. Within the different types of MaaS subsidies considered, subsidies aimed at sustainability exhibit a larger social valuation. Strong negative synergies among similar projects exist, signalizing that individuals prefer diversifying the use of public resources across different types of investment projects.

Optimizing the geometry of transportation networks in the presence of congestion.

Urban transport systems are gaining in importance, as an increasing share of the global population lives in cities and mobility-based carbon emissions must be reduced to mitigate climate change and improve air quality and citizens' health. As a result, public transport systems are prone to congestion, raising the question of how to optimize them to cope with this challenge. In this paper, we analyze the optimal design of urban transport networks to minimize the average travel time in monocentric as well as in polycentric cities. We suggest an elementary model for congestion and introduce a numerical method to determine the optimal shape among a set of predefined geometries considering different models for the behavior of individual travelers. We map out the optimal shape of fundamental network geometries with a focus on the impact of congestion.

Artificial intelligence models for pre-travel consultation and advice: yea or nay?

We assessed the accuracy of advice provided by ChatGPT for pre-travel counselling. ChatGPT appears to be complementary to the physical pre-travel consultation, giving appropriate advice and raising awareness. However, the advice lacked specificity, and travel physicians can help to fine-tune the advice provided to an individual's specific travel needs.

Discovering urban mobility structure: a spatio-temporal representational learning approach

ABSTRACTThe urban mobility structure is a summary of individual movement patterns and the interaction between persons and the urban environment, which is extremely important for urban management and public transportation route planning. The majority of current research on urban mobility structure discovery utilizes the urban environment as a static network to detect the relationship between people groups and urban areas, ignoring the vital problem of how individuals affect urban mobility structure dynamically. In this paper, we propose a spatio-temporal representational learning method based on reinforcement learning for discovering urban mobility structures, in which the model can effectively consider the interaction knowledge graph of individuals with stations while accounting for the spatio-temporal heterogeneity of individual travel. The experimental results demonstrate the advantages of individual travel-based urban mobility structure discovery research in describing the interaction between individuals and urban areas, which can account for the intrinsic influence more thoroughly.

When Will We Arrive? A Novel Multi-Task Spatio-Temporal Attention Network Based on Individual Preference for Estimating Travel Time

Predicting how long a trip will take may allow travelers plan ahead, save money, and avoid traffic congestion. The journey time estimation model should take into account three crucial factors: (1) individual travel preference, (2) dynamic spatio-temporal correlations, and (3) the association between long-term speed forecast and travel time estimate. In order to overcome these challenges, this study proposes a unique parallel architecture called the multi-task spatio-temporal attention network (MT-STAN) to estimate journey times. To extract the dynamic spatio-temporal correlations of the road network, we first develop a traffic speed prediction model based on spatio-temporal block and bridge transformer networks, combining the road, timestamp, and traffic speed information into hidden states. Second, we offer a personalized model for estimating journey times that makes use of cross-network, holistic attention, and semantic transformer. In this approach, travel preferences extraction through cross-network, holistic attention permits correlations between the dynamic road network’s hidden states and individual journey characteristics, which are subsequently transformed into global semantics by the semantic transformer; preferences and semantics are integrated during the estimate phase. Finally, a multi-task learning component is included, which combines both traffic speed prediction and individual journey time estimate, via the sharing of underlying network parameters and the improvement of the contextual semantic knowledge of the latter job. Evaluation experiments are carried out using a highway dataset collected in Yinchuan City, Ningxia Province, China. The proposed prediction model outperforms state-of-the-art baseline approaches in experiments.

Low-Carbon Travel Behavior in Daily Residence and Tourism Destination: Based on TPB-ABC Integrated Model

Low-carbon travel is considered as one of the most important strategies to reduce transportation carbon emissions, and its success is decided by the active participation of residents. Based on the theory of planned behavior (TPB) and Attitude-Behavior-Context theory (ABC), this study explores the influencing factors and formation paths of individual low-carbon travel behavior, and analyzes low-carbon travel behavior regarding both daily commuting from residence and tourism destinations. This study collects a sample of 506 respondents and uses Mplus 8.0 to examine the hypotheses. Empirical research results indicate that: (1) A certain gap exists in the individuals’ low-carbon travel behavior between daily residence and tourism destination. Differences exist in direct effects, mediating effects and moderating effects. (2) Low-carbon travel behavioral intention plays a significant mediating role in both daily residence and tourism destination, especially the former. Regarding daily residence, individuals’ attitude, subjective norms and perceived behavioral control have a positive effect on behavior through behavioral intention. Regarding tourism destination, only the attitude-low-carbon travel behavioral intention-behavior path is significant. (3) Situational factors play a significant positive moderating effect on the relationship between low-carbon travel behavioral intention and behavior, especially in tourism destination. This study reveals the internal mechanism of individuals’ low-carbon travel behavior and the differences between travel in daily life and tourism, helping to deepen understanding of individuals’ low-carbon travel behavior and providing guidance for promoting individuals’ low-carbon travel.

Batch-based vehicle tracking in smart cities: A Data fusion and information integration approach

A data fusion and information integration (DFII-VT) framework is proposed to solve the batch-based vehicle matching/tracking problem using heterogeneous fixed-location and mobile sensing data available in smart cities. To make the model more realistic, traffic knowledge such as lane-choice decision, traffic merging, travel time and vehicle characteristics are calibrated using the historical dataset and then integrated into the model. The problem can be formulated as a combinatorial optimization model, and solved using a dynamic programming/Kuhn-Munkres algorithm-based two-step approach. By doing so, the proposed method can obtain individual travel times of the matched data pairs, and these can be directly used to estimate the corridor travel times of individual vehicles. The experimental results show that the fusion of mobile sensing data and fixed-location data yields significantly better results than using single-source data. Significant improvements in matching accuracy are obtained as more traffic information is integrated into the model. Therefore, the method may be considered a framework for integrating the manifold traffic information acquired within smart cities to obtain more accurate matching results and further used to optimize other fine-grained traffic applications, such as estimating vehicle trajectories along arterial corridors, estimating individual vehicle-based fuel consumption/emissions, and helping to infer real-time queuing processes at signalized intersections. The paper also studies some practical issues related to the use of heterogeneous traffic data, such as data errors, and detection failures.

The Steller's Sea‐Eagle in North America: An economic assessment of birdwatchers travelling to see a vagrant raptor

Abstract Birdwatching—a cultural ecosystem service—is among the most popular outdoor recreational activities. Existing economic valuations of birdwatching typically overlook the economic contributions of birdwatchers travelling to see vagrant (out‐of‐range) birds. Economic valuations of vagrant birdwatching are few, and to date, no valuation of a large, charismatic vagrant species—or of a recurring individual vagrant bird—has been reported. During 2020–2022, a vagrant Steller's Sea‐Eagle (Haliaeetus pelagicus) was reported in several locations in North America, representing the first record of this species in these locations. In Winter 2020–2021, the eagle spent nearly a month on the eastern seaboard of the United States, and thousands of people travelled to see it. We conducted an online survey of individuals who travelled to see the eagle to estimate the individual and collective non‐consumptive use value of this vagrant birdwatching event. Using individual travel cost methodology, we estimated an average individual expenditure and, together with estimates of the total number of birdwatchers, we estimated non‐consumptive use value of the vagrant birdwatching event. Finally, we used a willingness to pay framework (via hypothetical donations to view the eagle) to evaluate the non‐consumptive use consumer surplus of the event. We estimated that, on average, individual birdwatchers spent $180 USD (95% CI = $156–$207) ignoring travel time—or $277 (95% CI = $243–$314) when accounting for travel time—to view the eagle. Furthermore, we estimated between 2115 and 2645 individuals travelled to see the eagle during December 2021 to January 2022. Thus, we estimated that the eagle generated a total expenditure between $380,604 and $476,626, or between $584,373 and $731,809 when accounting for travel time. Finally, based on travellers' willingness to pay, we estimated a non‐consumptive use consumer surplus of the event between $139,036 and $174,114. Assigning economic value to nature gives policymakers and business leaders footing to advance conservation in decision‐making. Although often overlooked in these decisions, vagrant birds supply ephemeral ecosystem services that might bolster community development efforts, particularly if vagrancy events occur with some predictability (e.g. recurring annually). Read the free Plain Language Summary for this article on the Journal blog.

Analysis of Disaster Risk on Expressways for Foreign Tourists to Japan Using Rental Cars: Study Using Central Nippon Expressway Pass

The travel patterns of foreign visitors to Japan are changing from group travel, which was the mainstream to date, to individual travel, and there has been an increase in the number of cars rented by visitors as a means of transportation. The use of rental cars improves access to rural areas and is being promoted as a means of transportation. However, if a natural disaster occurs while driving a rental car, then the individual may face difficulties owing to a lack of local support. Therefore, the traffic volume on the roads used by foreign visitors to Japan during certain periods needs to be estimated in advance as well as the natural disaster risks that exist on those roads. In this study, we utilized CEP data managed by the Central Nippon Expressway Company in Japan to identify the risk of disaster for foreign visitors to Japan who use expressways with rental cars. The CEP data has usage records of foreign visitors to Japan who used rental cars on expressways in the Chubu region of Japan during the period from 1 April 2018 to 31 March 2019. Using this data, we calculated the risk of earthquake and landslide disasters. Result of this analysis, it is clear that the risk of earthquake disasters is high on expressways in the Tokai region, and the risk of landslide disasters is high on expressways in mountainous areas.

Network equilibrium of battery electric vehicles considering drivers’ resting behavior

Driving fatigue cost is a major component of vehicle drivers’ travel costs in an intercity or regional network. The charging behavior of electric vehicle (EV) drivers, which is generally synchronized with drivers’ resting behavior, can contribute to the mitigation of drivers’ fatigue, especially after a prolonged driving period. Overlooking the impact of driving fatigue on the travel cost may overestimate the side-effects of the charging behavior for EV drivers, and result in biased flow and charging demand distribution. In this study, by considering the fatigue as part of the travel cost, we make the first attempt to investigate the impact of EV drivers’ charging and resting behaviors on their fatigue cost, and thus their travel plans and resultant flow distribution across the network. To this end, a novel network equilibrium modeling framework is first developed to capture the interaction among EV drivers’ travel plans, which specify the routing, recharging, and resting plans on a general intercity or regional network where charging stations and rest stops are deployed. When traveling between their origins and destinations, EV drivers are assumed to determine their travel plans to minimize their individual travel cost composed of driving time, rest time, charging cost, and fatigue cost, while preventing their batteries from being exhausted. The equilibrium model is then formulated as a variational inequality and transformed into a nonlinear optimization problem. An efficient solution algorithm integrating column generation and Benders decomposition approach is proposed to solve the established optimization problem. Numerical examples are presented to demonstrate the performance of the proposed models and solution algorithm. Numerical results validate that considering the impact of driving fatigue on the travel cost emphasizes the need for en-route charging for EV drivers with long-distance trips, and has an appreciable impact on the network flow and charging demand distribution. In addition, large-sized batteries and fast chargers may not necessarily reduce drivers’ travel costs for long-distance travel since charging behavior can be synchronized with drivers’ resting behavior and thus contribute to the mitigation of drivers’ fatigue.

Exploring willingness to use shared autonomous vehicles

Although multiple studies have modeled and predicted the potential effects of shared autonomous vehicles (SAVs), the research on the adoption of SAVs by riders with actual ridership experience is still limited. In addition, the increasing tendency towards operating SAV technology requires understanding its efficiency while integrating it into the existing transportation network infrastructure. This study aims to identify the factors affecting the user's willingness to ride the SAVs based on the data collected from a comprehensive survey distributed among users and non-users of a self-driving pilot project called RAPID (Rideshare, Automation, and Payment Integration Demonstration) in Arlington, Texas. Using structural equation modeling (SEM), we identify the effects from vehicle ownership, RAPID usage, existing modes of transportation, RAPID service attributes (comfort and safety), and sociodemographic variables on individuals' willingness to use SAVs in the future. Results indicate that most riders of the RAPID service are young Asian individuals and students from low-income households with limited or no access to a private vehicle. Furthermore, SEM results show that RAPID usage directly impacts willingness to use SAVs, implying that people start developing trust for the technology with an increase in the frequency of using the service. Our model suggests that sociodemographic attributes of the SAV riders indirectly influence the willingness to use SAVs through the mediators, including RAPID usage, existing modes of transportation, and vehicle ownership. This study provides crucial insights about individual travel behavior after integrating SAVs into existing transportation infrastructure to assist policymakers and transportation planners in developing AV-related policies.

Unpacking the Plus-Size Travel Movement

Efforts to promote inclusivity of marginalized groups in tourism have mainly focused on people with disabilities, seniors, and ethnic minorities. However, other marginalized groups such as plus-size travelers have been neglected. The Plus-Size Travel Movement draws attention to travel restrictions and ostracization from the market due to body size and intersectional identities. This study is the first of its kind to address this phenomenon. It aims to unpack the plus-size travel movement through a netnographic analysis of four major online movement platforms. The analysis reveals how community leaders mobilize with different discourses to politicize travel for plus-sized individuals and empower the community. This research seeks to stimulate further scholarly discussion about expanding theoretical approaches that examine consumer marginalization in tourism consumption beyond its current scope. It also provides recommendations for the industry, including engaging with plus-size travel influencers to establish more inclusive practices and policies and following the movement on social media closely to identify ways to enhance the experience of plus-size travelers.

The importance of utilizing travel history metadata for informative phylogeographical inferences: a case study of early SARS-CoV-2 introductions into Australia.

Inferring the spatiotemporal spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) via Bayesian phylogeography has been complicated by the overwhelming sampling bias present in the global genomic dataset. Previous work has demonstrated the utility of metadata in addressing this bias. Specifically, the inclusion of recent travel history of SARS-CoV-2-positive individuals into extended phylogeographical models has demonstrated increased accuracy of estimates, along with proposing alternative hypotheses that were not apparent using only genomic and geographical data. However, as the availability of comprehensive epidemiological metadata is limited, many of the current estimates rely on sequence data and basic metadata (i.e. sample date and location). As the bias within the SARS-CoV-2 sequence dataset is extensive, the degree to which we can rely on results drawn from standard phylogeographical models (i.e. discrete trait analysis) that lack integrated metadata is of great concern. This is particularly important when estimates influence and inform public health policy. We compared results generated from the same dataset, using two discrete phylogeographical models: one including travel history metadata and one without. We utilized sequences from Victoria, Australia, in this case study for two unique properties. Firstly, the high proportion of cases sequenced throughout 2020 within Victoria and the rest of Australia. Secondly, individual travel history was collected from returning travellers in Victoria during the first wave (January to May) of the coronavirus disease 2019 (COVID-19) pandemic. We found that the implementation of individual travel history was essential for the estimation of SARS-CoV-2 movement via discrete phylogeography models. Without the additional information provided by the travel history metadata, the discrete trait analysis could not be fit to the data due to numerical instability. We also suggest that during the first wave of the COVID-19 pandemic in Australia, the primary driving force behind the spread of SARS-CoV-2 was viral importation from international locations. This case study demonstrates the necessity of robust genomic datasets supplemented with epidemiological metadata for generating accurate estimates from phylogeographical models in datasets that have significant sampling bias. For future work, we recommend the collection of metadata in conjunction with genomic data. Furthermore, we highlight the risk of applying phylogeographical models to biased datasets without incorporating appropriate metadata, especially when estimates influence public health policy decision making.

A Spatial-Temporal Graph Convolutional Recurrent Network for Transportation Flow Estimation.

Accurate estimation of transportation flow is a challenging task in Intelligent Transportation Systems (ITS). Transporting data with dynamic spatial-temporal dependencies elevates transportation flow forecasting to a significant issue for operational planning, managing passenger flow, and arranging for individual travel in a smart city. The task is challenging due to the composite spatial dependency on transportation networks and the non-linear temporal dynamics with mobility conditions changing over time. To address these challenges, we propose a Spatial-Temporal Graph Convolutional Recurrent Network (ST-GCRN) that learns from both the spatial stations network data and time series of historical mobility changes in order to estimate transportation flow at a future time. The model is based on Graph Convolutional Networks (GCN) and Long Short-Term Memory (LSTM) in order to further improve the accuracy of transportation flow estimation. Extensive experiments on two real-world datasets of transportation flow, New York bike-sharing system and Hangzhou metro system, prove the effectiveness of the proposed model. Compared to the current state-of-the-art baselines, it decreases the estimation error by 98% in the metro system and 63% in the bike-sharing system.

Influence of residential built environment on human mobility in Xining: A mobile phone data perspective

The emergence of large-scale spatiotemporal trajectory data offers an excellent opportunity to characterize collective human mobility patterns and their relationship with the urban built environment. Such an approach can generate a complementary understanding of traditional individual travel behavior based on travel survey data. Using mobile phone data, this study aims to investigate how residential built environments affect residents’ mobility at an aggregated level. Specifically, three indicators (movement distance, activity space, and the number of stops) were derived from raw mobile phone data to characterize human mobility. The decay coefficients and average values were then employed to reveal the aggregated characteristics of movement distance, activity space, and the number of stops. Furthermore, linear regression was applied to examine the relationship between human mobility indicators and residential built environments. The results indicate that the living-built environment could better explain the activity space than the movement distance and number of stops. In addition, some differences in the relationship between human mobility and residential built environments are identified between weekdays and weekends. These findings could provide new insights into human mobility and its interaction with the built environment, thus advancing our understanding of human travel behavior from both individual and collective perspectives.

A latent auction model of firm establishment location choice in the Greater Toronto Area

We present a maximum bid model for business establishments in the Greater Toronto Area (GTA). The model is based on a latent auction approach, wherein price and utility are jointly modeled in a random utility-maximizing framework. Structural relationships between establishments by their industry classification and type (i.e., headquarters, subsidiary, branch, or single location) are explored through a generalized nesting correlation structure. This structure provides a mechanism for differentiating between branch and single-location establishments, having seemingly similar functions, but differing in their allocation between industry and type nests. Maximum bid models are estimated for the full establishment population in the GTA at the level of individual buildings. The model scope allows us to provide a range of insights into establishment location choice behavior, which was previously infeasible due to data limitations. We find that professional service establishments tend to locate near passenger rail stations (i.e., subway and LRT), while industrial establishments tend to locate near major highway interchanges. We confirm the finding by Kenworthy and Laube (1999) that Canadian cities are less auto-dependent than US cities, albeit from the perspective of establishment location rather than individual vehicle travel data.

A Path Recommendation Method Considering Individual Driving Preferences

The issue of congestion on urban roads stems from an imbalance between transport demand and supply. It has become imperative to address the problem from the traffic demand side. While managing effective traffic demand relies on understanding the individual preferences of drivers, the current method for gathering preferences (i.e., through questionnaires) is both expensive and may not accurately capture the characteristics of respondents due to their varying interpretations of the options. To overcome these challenges, we proposed a path recommendation method that takes individual travel preferences into consideration by employing automatic license plate recognition (ALPR) data for the extraction of individual travel preferences. We initially identified key factors influencing the path selection behaviors of drivers, including path attributes, travel attributes, and individual attributes. Subsequently, we constructed a path satisfaction model based on individual preferences, employing an improved analytic hierarchy process (AHP). Furthermore, we utilized the pth percentile approach, rather than expert scores, in order to determine the relative importance of each indicator in the improved AHP. By applying the proposed model to the ALPR data from Xuancheng City, we successfully extracted the path selection preferences of drivers. We designed various scenarios to verify the reliability of the model, and the experimental results demonstrated that the proposed path satisfaction model can effectively capture the influence of underlying indicators on the path selection behavior of individuals with diverse travel preferences, considering different driver types and path attributes. Moreover, compared to the real trajectory, the recommended paths yielded an overall satisfaction improvement of over 10%, confirming the reliability and practicality of our proposed model.