Search

Validation of travel demand models, although recognised as important, is seldom undertaken. This study adds to the scarce literature in this field by undertaking an external validation of a multi-modal transit route choice model. The model was estimated using smart card data for the urban transit network of Amsterdam before the introduction of a new metro line and is used to predict changes in travel behaviour after the network change. To validate, the model was checked for changes in estimated parameters between the two time periods, and predictive ability was evaluated at different aggregation levels. Although most model parameters were found to be unstable between the two contexts, the predictive performance at all levels was similar to the locally estimated model. Moreover, individual choices and transit mode-share predictions were found to be close to the observed ones. The errors were relatively larger for the link and route-level predictions, some of which could be attributed to the assumptions made regarding consideration choice set given as input to the model. On comparing alternative model specifications, using generic instead of mode-specific travel attributes lead to a strong degradation in predictive performance. Conversely, a model incorporating overlap between routes, with a better model fit in the base period, did not offer a clear improvement in prediction performance. The study highlights the need to validate transit route choice models before using them for deriving policy recommendations, especially in this data-rich age in which it can often be undertaken at a relatively low additional cost.

With the arrival of new technologies more en-route traffic information sources have become available, especially in-car information sources. The aim of this study is to gain more insight into the effect of multiple, and possibly conflicting, sources of information on route choice and driver behaviour. In a driving simulator experiment, participants were required to make multiple drives, each of which ended with a choice between the normal and an alternative route. On each trial participants received traffic information from a Variable Message Sign (VMS), i.e. a dynamic sign above the road providing descriptive traffic information in the form of expected travel times (ETTs), a navigation device providing in-car prescriptive route advice, or information from both sources. In the latter type of trial the information could be congruent or conflicting with regards to ETTs on the VMS and advise from the navigation. After each trial, participants indicated how much trust they had in the traffic information and their primary information source. A Bayesian model was used to quantify the propensity to switch to the alternative route. Results indicate that overall compliance was very high for the primary source even when the other source did not corroborate this information and that most participants preferred to use the information from a VMS. However, when both the VMS and the navigation device provided information and the VMS indicated the same ETTs for the normal and alternative route, route choice was influenced by the advice provided by the navigation device. Also, in this type of trial mean speed was significantly lower compared to trials in which the two sources were in conflict, indicating increased mental workload, most likely due to attentional dissonance: a situation in which stimuli compete for attention resulting in cognitive conflict and the need to inhibit non-relevant information. A deeper understanding of how drivers use multiple traffic information sources and cope with irrelevant information could support driver safety and comfort, increase the usability of information sources, and help reduce stress, anxiety, and information overload while driving.

We propose a route choice model in which traveler behavior is represented as a utility maximizing assignment of flow across an entire network under a flow conservation constraint. Substitution between routes depends on how much they overlap. The model is estimated considering the full set of route alternatives, and no choice set generation is required. Nevertheless, estimation requires only linear regression and is very fast. Predictions from the model can be computed using convex optimization, and computation is straightforward even for large networks. We estimate and validate the model using a large dataset comprising 1,337,096 GPS traces of trips in the Greater Copenhagen road network. We propose a route choice model in which traveler behavior is represented as a utility maximizing assignment of flow across an entire network under a flow conservation constraint. Substitution between routes depends on how much they overlap. The model is estimated considering the full set of route alternatives, and no choice set generation is required. Nevertheless, estimation requires only linear regression and is very fast. Predictions from the model can be computed using convex optimization, and computation is straightforward even for large networks. We estimate and validate the model using a large dataset comprising 1,337,096 GPS traces of trips in the Greater Copenhagen road network.

The tourism industry is rapidly growing, and the massification of certain areas is jeopardising the environment. Certain roads in protected areas or near tourist attractions are experiencing an increase in traffic volumes, leading to higher pollution and noise levels and greater discomfort in the travel experience. To recommend measures intended to ensure a sustainable tourism industry without compromising the environment, it is necessary to obtain further knowledge regarding tourists' travel behaviour as it may differ from that of other travellers, i.e., tourists might choose a route based on variables other than time and cost, such as landscape or tourist attractions. Using Geiranger Fjord in Norway as a case study, tourist route choice preferences were observed and analysed. In total, 60 routes from 12 different origins were observed based on 408 responses collected through a tourist survey administered in the summer of 2018. These data were combined with information regarding road features and the locations of tourist attractions. A path size correction logit model was selected to estimate the significant variables in the choice of one route over other routes. The results showed that tourists selected their routes mainly based on the travel time, road scenery (water bodies and forests), sightseeing places among first-time visitors, and outdoor activities among those living outside of Norway. In addition, the tourists preferred local roads, and 'caravan' tourists were more attracted to roads with fewer facilities. The findings of this study show that tourists’ route choice behaviour differs from that of other travellers. Policymakers should consider these differences to enhance transport systems. Further research should focus on incorporating these findings into a transport model to simulate tourists' travel patterns and quantify changes in traffic volumes due to different sustainable measures.

This study proposes the use of multi-criteria decision models (MCDM) for transportation mode-route choice decisions. This method is beneficial when trips' microdata are unavailable. Route-mode choice decisions were investigated for three public transportation modes (buses, railways, and airlines) in the post-China Pakistan Economic Corridor (CPEC) investment in Pakistan Railways (PR) for a link between Peshawar and Karachi. TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) was used for the mode choice decisions and a hybrid model of AHP (Analytical Hierarchy Approach) – TOPSIS was used for the route choice decision ML-1 link of PR. This study concludes that rails were the best mode of transportation in post-CPEC investment. Furthermore, route 3, linking Karachi to Peshawar via Lodhran, Multan, and Miniawali, is the best route connection among the four considered routes.

Optimizing route choices for truck drivers is a key element in achieving reliable road freight operations. For commercial reasons, it is often difficult to collect freight activity data through traditional surveys. Automated vehicle identification (AVI) data on fixed locations (e.g., Bluetooth or camera) are low-cost alternatives that may have the potential to estimate route choice models. However, in cases where these AVI sensors are sparsely located, the resulting data lack actual route choices (or labels), which limits their application estimating route choice models. This paper overcomes this limitation with a new two-step approach based on fusing AVI and loop-detector data. First, a sparse Bluetooth data set is fused with travel times estimated from densely spaced loop-detector data. Second, the combined data set is fed into a bi-objective optimization method which simultaneously infers the actual route choices of truck drivers between an origin–destination pair and estimates the parameters of a route choice (discrete choice-based) model. We apply this approach to investigate the route choice behavior of truck drivers operating to and from the port of Rotterdam in the Netherlands. The proposed model can distinguish between peak and off-peak periods and identify different segments of truck drivers based on a latent classes choice analysis. Our results indicate the potential of traffic and logistics interventions in improving the route choices of truck drivers during peak hours. Overall, this paper demonstrates that it might be possible to estimate route choice characteristics from readily available data that can be retrieved from traffic management agencies.

Pedestrians often need to decide between different routes they can use to reach their intended destinations, both during emergencies and in their daily lives. This route-choice behavior is important in determining traffic management, evacuation efficiency and building design. Here, we use field observations and a virtual experiment to study the route choice behavior of pedestrians around obstacles delimiting exit routes and examine the influence of three factors, namely the local distance to route starting points and the pedestrian density and walking speeds along routes. Crucially, both field study and virtual experiment consider the same scenario which allows us to directly assess the validity of testing pedestrian behavior in virtual environments. We find that in both data sets the proportion of people who choose a closer exit route increases as the difference in distance between exit route starting points increases. Pedestrians’ choices in our data also depend on pedestrian density along routes, with people preferring less used routes. Our results thus confirm previously established route choice mechanisms and we can predict over 74% of choices based on these factors. The qualitative agreement in results between the field study and the virtual experiment suggests that in simple route-choice scenarios, such as the one we investigate here, virtual experiments can be a valid experimental technique for studying pedestrian behavior. We therefore provide much-needed empirical support for the emerging paradigm of experiments in virtual environments.

Suppose someone wants to walk from one location to another location. In doing so, they either follow a pre-planned route or decide which way to go each time they reach a junction. Research into pedestrian dynamics has carefully studied route planning and single instances of decisions on routes. However, little is known about the decision-making processes in sequences of consecutive pedestrian route choices. Here, we propose the first mathematical model for this situation. Its key hypothesis is that the sensitivity of pedestrians to environmental information, such as signs or the movement of others, diminishes, the more decisions they make in sequence. To validate our model, we conduct a virtual reality experiment with over 200 participants. Our findings suggest that sensitivity to information diminishes for the experimental case when only information on the movement of others is available to pedestrians. Comparison of our model to situations when signs are present and to data from previously published work suggests that this effect cannot be detected in short sequences of route decisions that contain three or fewer decision points. We demonstrate the implications of this diminishing sensitivity to environmental information at the example of egress from a metro-station. While the proportion of pedestrians using a signposted shortest route may decrease, following behaviour is also suppressed which may result in more predictable route choice dynamics across pedestrian crowds. We advocate further study on this topic in real-world settings, to investigate the need for appropriate pedestrian route choice management strategies. Our experiment also provides preliminary insights into how different personality traits affect following behaviour and we suggest this opens another avenue for future investigation.

This study contributes to the field of cycling route choice by adopting the unprecedented combination of the Metropolis-Hastings (MH) path-sampling algorithm and the Expanded Path Size Logit (EPSL) model. The MH sampling approach is used to generate 15 alternative route choice sets for cyclists. The EPSL multivariate route choice framework is utilized to account for the correlation between sampled and non-sampled alternatives (joint MH-EPSL model). The data used in this paper is drawn from GPS data collected by the City of Toronto using a custom-built smartphone application in 2014–2015. The study focuses on non-work-related cycling trips (shopping, leisure, social and others) in downtown Toronto on weekdays.The estimated results indicate that the presence of bicycle lanes and road medians attractions and number of trees along the path have a positive impact on cyclist route choice. In general, cyclists prefer to take shorter routes on lower speed roads with less public transit stops especially during the evening rush hour, and less willing to take one-way streets, local roads, and steep road segments. These findings are useful to policy makers as well as transportation and urban designers when developing a cycling network aiming to attract more cyclists. Finally, our results indicate that the MH-EPSL model performance is an appropriate framework to study cyclists’ route choice decisions. Keywords: Route choice, GPS data, Expanded Path Size Logit, Metropolis-Hastings, Cyclists

Given the aim of increasing public transport patronage, it is important to understand how passengers perceive different trip characteristics. Most of the existing studies about public transport demand and route choice assigned a higher value of time to transfers than in-vehicle time and used a general transfer penalty to capture an average increase in the travel disutility because of the amount of transfers. However, it is likely that there are nuances to the transfer behaviour depending on specific transfer conditions that existing models do not capture and hence it is difficult to evaluate measures aimed at improving transfers to make public transport more attractive. This study presents a route choice model for the large-scale multimodal public transport network in the Greater Copenhagen Region where a variety of transfer attributes were explicitly considered within a unified model framework. The model was estimated on an extensive revealed dataset of 4,810 observed routes that made it possible to evaluate the rates of substitution of transfer related attributes. The results revealed that travellers do consider attributes for transfers such as ease of wayfinding, presence of shops and escalators at stations when choosing routes in the public transport network and this influences the attractiveness of the respective routes with a quite large range of the transfer penalty from 5.4 min compared to bus in-vehicle time for the best possible transfer to 12.1 min for the worst. Furthermore, the study revealed some differences in the preferences for transfer attributes across passengers. This suggest a quite large potential for improving transfers and hence public transport patronage focusing on the attributes of the transfers.

Path Size Logit route choice models attempt to capture the correlation between routes by including correction terms within the route utility functions. This provides a convenient closed-form solution for implementation in traffic network models. The path size terms measure distinctiveness of routes; a route is penalised based on the number of other routes sharing its links, and the costs of those shared links. Typically, real road networks have many very long routes that should be considered unrealistic. Such unrealistic routes are problematic for the Path Size Logit (PSL) model because they negatively impact the choice probabilities of realistic routes when links are shared. The Generalised Path Size Logit (GPSL) model attempts to address this problem by weighting the contributions of routes to path size terms according to the ratio of route travel costs. However, the GPSL model is not internally consistent in how it defines routes as being unrealistic: the path size terms consider only travel cost, whereas the route choice probability relation considers disutility including the correction term. To solve these challenges, this paper formulates a new internally consistent Adaptive Path Size Logit (APSL) model wherein routes contribute to path size terms according to the ratio of route choice probabilities, ensuring that routes defined as unrealistic by the path size terms, are exactly those with very low choice probabilities. The APSL route choice probability relation is an implicit function, naturally expressed as a fixed-point problem. A proof is provided for the guaranteed existence of solutions, as well as conditions for the uniqueness of solutions. A Maximum Likelihood Estimation procedure is given for estimating the APSL model with tracked route observation data, and this procedure is investigated in a simulation study where it is shown it is generally possible to reproduce assumed true parameters. APSL is then estimated using real tracked route GPS data on a large-scale network, and results are compared with other PSL models..

Travelers make different choices about their routes based on the perceived travel time. This study aims to analyze the route choice behaviors by grouping them according to their perceived travel times. Reliability preference is a behavioral characteristic for travelers’ on-time arrival according to stochastic travel time, and it is used to classify the travelers into four groups. From Dedicated Short Range Communication (DSRC) data in Daegu, 9,758 travelers are observed to analyze their route choice behaviors. A stochastic model with generated sets of choices is developed using the probabilistic characteristics of travel times. A route-choice model is also estimated for each group, and the travelers’ route-choice behaviors are analyzed using a mixed path-size correction logit (MPSCL) model. The statistical analysis indicates that there are different route-choice behaviors among the four groups. The different behaviors concerning the routes are analyzed by comparing the magnitude of the different coefficients. This study contributes to constructing different strategies for acquiring information considering the route-choice behaviors of each group. Providing travel information that considers individual preferences in travel choice is essential in smart mobility transportation systems.

The paper illustrates a route choice model based on the recently proposed Combination of Random Utility Models (CoRUM) model by Papola (2016). Specifically, the CoRUM specification with nested logit components, termed Combination of Nested Logits (CoNL), accounts for any correlation patterns amongst alternatives while keeping a closed-form expression for both probabilities and correlations. Thus, the paper illustrates a CoNL route choice model capable to target correlations between routes based on their topological overlapping, and characterized by a closed-form probability statement. The CoNL route choice model is operationalized by means of an algorithm providing automatically CoNL specification and corresponding route choice probabilities on a set of enumerated paths. Model performance is tested on various networks, with very satisfactory results.

The majority of existing navigation systems only account for a single aspect of the route choice, like travel time or distance, in finding the optimal route for the trips in the network. In this research, we first identify a range of diverse factors that travelers take into account in their route choice decision in the network. A stated preference survey is conducted to show the heterogeneity in the preferences of users and its dependence to the purpose of the trips over the weekdays and weekends. Interestingly, results of the survey show that road safety is the most influential factor in the route choice decision of the average participants over weekends, exceeding even the travel time, and participants give more importance to the scenic quality of the routes for their weekend trips in comparison to their weekday trips. The results of the second part of the survey also indicate that in 27% of the cases participants choose routes other than the ones suggested by navigation systems, and 33% of the times that they take the suggested routes, they modify these routes according to their own preferences. The partial inability of existing navigation systems to suggest the routes that match the preferences of users can be attributed to ignoring (1) the diversity in influential factors and (2) the heterogeneity in preferences of the users by these systems. We propose a dynamic mixed logit route choice model to include the effects of information and learning to estimate parameters of a multivariable utility function for individual users based on their own historical route choice data over time. Finally, we present the concept of a smart navigation system that can gather the required information from real-time and online sources to suggest the routes that best match the users’ own preferences.

The proliferation of information systems is enabling drivers to receive en route real-time travel information, often from multiple sources, for making informed routing decisions. A robust understanding of route choice behavior under information provision can be leveraged by traffic operators to design information and its delivery systems for managing network-wide traffic. However, most existing route choice models lack the ability to consider the latent cognitive effects of information on drivers and their implications on route choice decisions. This paper presents a hybrid route choice modeling framework that incorporates the latent cognitive effects of real-time information and the effects of several explanatory variables that can be measured directly (i.e., route characteristics, information characteristics, driver attributes, and situational factors). The latent cognitive effects are estimated by analyzing drivers’ physiological data (i.e., brain electrical activity patterns) measured using an electroencephalogram (EEG). Data was collected for 95 participants in driving simulator experiments designed to elicit realistic route choices using a network-level setup featuring routes with different characteristics (in terms of travel time and driving environment complexity) and dynamic ambient traffic. Averaged EEG band powers in multiple brain regions were used to extract two latent cognitive variables that capture driver’s cognitive effort during and immediately after the information provision, and cognitive inattention before implementing the route choice decision. A Multiple Indicators Multiple Causes model was used to test the effects of several explanatory factors on the latent cognitive variables, and their combined impacts on route choice decisions. The study results highlight the significant effects of driver attributes and information characteristics on latent cognitive effort and of route characteristics on latent cognitive inattention. They also indicate that drivers who are more attentive and exert more cognitive effort are more likely to switch from their current route by complying with the information provided. The study insights can aid traffic operators and information service providers to incorporate human factors and cognitive aspects while devising strategies for designing and disseminating real-time travel information to influence drivers’ route choices.

In transportation planning and design studies it is customary to consider the network in its equilibrium state. Equilibrium conditions in the presence of Advanced Traveller Information Systems (ATIS) and recurrent congestion have been formulated either according to some behavioural principle derived from Wardrop’s assumption, or as fixed-point states of day-to-day dynamic assignment processes. In the latter approach only, a differentiation has been introduced between predictive ATIS and ATIS providing static information, but the impacts on equilibrium travel times are relatively unexplored. In addition, the route choice updating model is formulated without consideration of state-dependent effects, such as inertia to change. This forms the motivation for the present paper. According to traffic information sources, three classes of users are considered: (a) users with predictive information, (b) users with static information and compliant, (c) users not equipped with ATIS or noncompliant. Users of the three classes make choices in a stochastic manner based on a logit model subject to inertia. Users of the first class have lower perception variance in view of the higher information quality. The compliance rate is endogenous and dependent on the information accuracy of the static ATIS. The fixed points in the class-specific route flows of the dynamic day-to-day processes characterise a new concept of network equilibrium, referred to as Mixed User Equilibrium (MUE), where, if each user shifts from her currently used route to her newly chosen route, the observed class-specific route flows do not change. A variant of the method of successive averages is proposed for computing MUE. The model is illustrated by an example related to the Nguyen-Dupuis network. The impacts on equilibrium travel times of the market penetration of ATIS with different functionalities as well as of inertia are examined.

Understanding the relationship between micro-scale built environment and pedestrian trips from/to metro stations is important to stimulate walking as the key egress/access mode. This study developed a path size correction latent class logit model using 515 observed route trajectories between a metro station and an origin/destination near the metro station in Tianjin, China. Results suggest that pedestrians can be classified into two latent classes. Pedestrians in both classes prefer routes with a fence and lower floor area ratio. Except for the common significant results in the two classes, pedestrians who work in the area in class 1 attach a higher utility to shorter routes, lower width of the sidewalks, less street greenery, less traffic lights number, and smaller width of the road. Pedestrians who do not work in the area in class 2 attach a higher utility to longer routes, higher building lot coverage, and larger width of the road.

As cities make concerted efforts to become more bicycle-friendly through policy changes and infrastructure, it is important that such efforts ultimately support people who currently bicycle and remove barriers that may prevent more people from bicycling. Travel surveys can reveal the nature and distribution of bicycling trips, but perceptions and behaviours of bicyclists are better understood through qualitative methods. In particular, developing cycling cities are unique settings to examine how the built environment supports bicycling as changes are made to increase bicycling levels. Through semi-structured interviews, this research explores the case of Hamilton, Ontario and the factors that influence route choice from the perspective of regular bicyclists. Major themes were identified using thematic analysis: (i) exclusion from road space; (ii) infrastructure; and (iii) streetscape. Bicyclists highly value infrastructure and seek routes that minimize interactions with cars, while avoiding many arterial roads that prioritize motorists. Routes that appear to be more human-oriented or that have nature are also attractive. Many regular bicyclists report that current bicycle infrastructure does not meet their preferences. Despite building nearly half of the planned infrastructure network, our findings suggest the built environment is not perceived to be oriented to bicycling. This study provides policy and practice recommendations for developing cycling cities in North America as they transition towards established cycling cities.

This paper discusses the use of smart card big data to understand and evaluate the public transport network in Seoul, South Korea. Smart card data (SCD) trap multiple transit modes and cover a larg...

Uncertainty in transportation systems can incur additional travel costs, but this adverse effect can be mitigated by providing travel information (Lindsey et al., 2014). It is not clear, however, if providing information always reduces system travel costs, particularly when the information provided is not one hundred percent accurate. This paper studies the welfare effects resulting from providing pre-trip information to morning commuters in a two-route network with bottlenecks, where the bottleneck capacity and free-flow travel time on each route are stochastic. We derive the expected travel costs at user equilibrium (UE) under stochastic conditions without and with pre-trip information first on one route, and later on two routes. Under this model, we examine how accurate and inaccurate information affect commuting costs. We find that full and accurate information is welfare-improving under stochastic bottleneck capacity and deterministic free-flow travel time. However, when bottleneck capacity and free-flow travel time are both stochastic, full and accurate information can be welfare-reducing for the two-route scenario. We show that the degree of correlation between routes in traffic conditions, the frequency and severity of bottleneck capacity drops, and the relationship between free-flow travel time and bottleneck capacity significantly affect the welfare effects of pre-trip information. Furthermore, when bottleneck capacity experiences severe drops, full information is more likely to be welfare-improving, even if the provided information is not completely accurate. These theoretical results are supplemented by numerical examples that show the welfare effects of providing pre-trip information to morning commuters.

Street attributes are thought to play an important role in influencing pedestrian route choices. Faced with alternatives, pedestrians have been observed to choose faster, safer, more comfortable, more interesting or more beautiful routes. Literature on pedestrian route choice has provided methods for assessing the likelihood of such options using discrete choice models. However, route choice estimation, which is data intensive and computationally challenging, remains infrequently deployed in planning mobility analysis practice. Even when coefficients from previous studies are available, operationalizing them in foot-traffic predictions has been rare due to uncertainty involved in the transferability of behavioral effects from one context to another, as well as computational challenges of predicting route choice with custom attributes. This paper explores a simpler method of route choice prediction, implemented in the Urban Network Analysis toolbox, which assigns probabilities to available route options based on distance alone. We compare the accuracy of distance-weighted approaches to the more detailed utility-weighted approach using a large dataset of observed GPS pedestrian traces that include numerous trips between same intersections pairs in downtown San Francisco as a benchmark. Even though a utility-weighted model matches observed pedestrian flows most accurately, a distance-weighted model is only marginally inferior, on average. However, shortest-distance and highest-utility route predictions are both significantly inferior to the utility-weighted and distance-weighted sample-enumeration methods. Our findings suggest that simplified assumptions can be used to predict pedestrian flow in practice with existing software, opening pedestrian flow predictions to a wider range of planning and transportation applications.

Travel time reliability plays an important role in travelers’ route choice behaviors. Based on a previously developed generalized Bayesian traffic model, we propose different types of perceived knowledge (i.e., mean-variance-based type, relative gap-based type, and penalty-based type) to model travelers’ daily route choice behavior concerning travel time reliability. We theoretically demonstrate the flexibility of the generalized Bayesian model in capturing various existing UE-based travel behaviors and other non-UE-based travel behaviors (e.g., penalty-based) in stochastic transportation systems. Three major conclusions are obtained. First, the route choice dynamics induced by the Bayesian model with an infinitely long memory and mean-variance-based perceived knowledge will converge to the mean-variance UE condition. Second, the convergence of route choice dynamics to a UE condition is not affected by adding a bounded weight on the daily perceived knowledge. Thirdly, non-UE-based formulations of perceived knowledge also lead to fixed points for the mean route choice proportion. The convergences of the models with different types of perceived knowledge are verified based on numerical studies and the underlying day-to-day route choice dynamics with both recurrent and non-recurrent unreliability are examined.

While big data helps improve decision-making and model developments, it often runs into privacy concerns. An example would be retrieving drivers’ origin and destination information from smartphone navigation apps for developing a route choice behavior model. To conserve privacy, yet to take advantage of big data in navigation applications, the authors propose to apply a federated learning approach, which has shown promising application in predicting smartphone keyboard’s next word without sending text to the server. Additional benefits of using federated learning is to save on data communications, by sending model parameters instead of entire raw data, and to distribute the computational burden to each smartphone instead of to the main server. The results from real-world route navigation usage data from about 30,000 drivers over one year showed that the proposed federated learning approach was able to achieve very similar accuracy to the traditional centralized global model and yet assures privacy.

Understanding the factors influencing bicycling is important to improve educational and built-environment investments to increase bicycling rates. Although factors such as the physical environment, sociodemographics, and psychology influence bicycling, safety is also one of the primary reasons people avoid bicycling. Interventions based on objective safety can reduce the number of crashes, injuries, and fatalities, but people might still feel uncomfortable bicycling owing to subjective safety (perceived safety or avoidance of risk). Several studies have examined the subjective safety of bicyclists based on stated-preference surveys, but these studies have limitations, including response bias. A revealed preference method was implemented by combining 9,101 bicycling trips of Grid Bike Share in Arizona with transportation network and crash data. A segmented path size correction logit model identified that regular bicyclists took a 1.6-times longer detour to avoid historic crash locations than casual bicyclists. These two groups also exhibited different choices related to the built environment and navigation. The significance of different types of bicyclists avoiding historic crash locations or risky infrastructure is that this indicates that crash datasets coupled with route data could be used as one of several indicators of perceived safety. Recommendations to increase perceived safety and reduce the crash risk of occasional bicyclists include expanding bicycle-specific infrastructure, constructing contraflow bicycle lanes in a one-way street, separating high volume lanes with a bike lane, and improving the education of road users.

Route choice is often modelled as a two-step procedure in which travellers choose their routes from small sets of promising candidates. Many methods developed to identify such choice sets rely on assumptions about the mechanisms behind the route choice and require corresponding data sets. Furthermore, existing approaches often involve considerable complexity or perform many repeated shortest path queries. This makes it difficult to apply these methods in comprehensive models with numerous origin-destination pairs. In this paper, we address these issues by developing an algorithm that efficiently identifies locally optimal routes. Such paths arise from travellers acting rationally on local scales, whereas unknown factors may affect the routes on larger scales. Though methods identifying locally optimal routes are available already, these algorithms rely on approximations and return only few, heuristically chosen paths for specific origin-destination pairs. This conflicts with the demands of route choice models, where an exhaustive search for many origins and destinations would be necessary. We therefore extend existing algorithms to return (almost) all admissible paths between a large number of origin-destination pairs. We test our algorithm on a road network modelling the Canadian province British Columbia and analyze the distribution of locally optimal paths in the province., Comment: Keywords: alternative paths; choice set; local optimality; road network; route choice

More than seven million people rely on the metro system daily in Seoul, South Korea. The metro system plays a vital role in connecting passengers to desired locations with the advantage of allowing for travel reliability and relative safety, all the while being an environmentally friendly alternative to other transport systems. Despite the benefits mentioned above, crowding on the metro can contribute significantly to deterioration of the individual’s travel experience and necessitates its proper evaluation, so that user experience may be improved. To achieve this, this study quantifies the level of crowding by deriving estimates for route choice models with crowding as one of the main attributes for metro passenger decisions. The data used in the model are obtained through a stated preference survey, which was conducted at five major transfer stations in Seoul to estimate factors such as travel time, transfers, travel cost, and crowding level. These attributes are then analyzed using homogeneity and heterogeneity models of different trip purposes. Furthermore, crowding multipliers (CMs) for Seoul are estimated and compared with those of other cities around the world. Estimation results pinpoint crowding as one of the main factors in route choice for passengers for all types of trips, with multiplier values reaching the highest of all cities around the world for sitting CM and standing CM at 2.15 and 3.22, respectively. Our results indicate that Seoul metro passengers are more sensitive to crowding than any of the passengers analyzed in 18 major city metros around the world.

Background Variable message signs (VMS) have been widely applied to alleviate traffic congestion by providing spatiotemporal information related to traffic and network conditions to drivers. There is a debate on whether the VMS information would lead to the less-alleviation of congestion or congestion transfer to the adjacent roads in the different traffic congestion situations due to the inappropriate VMS information. Additionally, it is still unknown how to find a practically applicable method to design the most effective VMS information for drivers. Objective In order to design and display the appropriate VMS information for drivers under different traffic conditions, this paper aims to identify the effects of VMS on individuals’ route choice behavior by exploring the different effects of VMS on the several separate groups. Drivers in these groups are homogenous in gender, driving experience, education level, occupation, and monthly income. Methods In the hypothetical scenario, a stated preference (SP) questionnaire was conducted to build the dataset consisting of 3462 samples from local private and taxi drivers. The decision tree via the chi-squared automatic interactive detector (CHAID) algorithm is utilized to split drivers into several groups, and further two fixed-parameter logistic regression models considering intra-heterogeneity between drivers (i.e., considering and not considering systematic taste variations), and random-parameter logistic regression models are employed to analyze the factors affecting driver route choice behavior of each group. Results Five groups with different driver characteristics are identified. Besides, the modeling results indicate that the effect of the VMS information on individual route choice behavior is heterogeneous. These findings include: (i) random-parameter logistic regression models have a significantly improved goodness-of-fit because of the lower AIC and BIC; (ii) compared with the stated cause information on road construction and congestion, the accident information had a larger effect on driver route choice behaviors; (iii) the route choice behaviors of drivers with more than 6-year driving experience are affected by the VMS information on road constructions, accidents, and the quantitative delay time; (V) the route choice behaviors of less-experienced female drivers were affected by display formats and display frequencies of VMS information. Conclusions The findings from this study can assist traffic authorities in designing the most appropriate VMS for different traffic congestions based on driver characteristic distribution and the road capacity for improving the practicability of VMS information and further provide the theoretical evidence for the design of in-vehicle personalized information service systems.

Driving in congested traffic is a nuisance that not only results in longer travel times, but also triggers frustration and impatience among drivers. A few studies have modeled the effects of congested traffic in the resulting route choice behavior of car drivers. The studies used frequentist models such as discrete choice models to analyze large samples. However, these studies did not compare the inferences obtained from the frequentist and Bayesian approaches, particularly for datasets which are not sufficiently large. It has been shown by researchers that Bayesian models perform well, especially when the sample size is small. Thus, this paper develops and compares a multinomial logit (frequentist) and a Naïve Bayes (Bayesian) model on a mid-sized dataset of size around 100 participants which was obtained from a driving simulator experiment to understand driver’s route choice under stop-and-go traffic. The results show that the prediction power of the Naïve Bayes model is much higher than the multinomial logit model (MNL). The Naïve Bayes model is also found to perform better than machine learning algorithms like the decision tree model. The findings from this study will be useful to researchers and practitioners as they should test both the approaches and select the appropriate model, particularly in the case of seemingly large datasets.

Route choice modeling is a fundamental task in transportation planning and demand forecasting. Classical methods generally adopt the discrete choice model (DCM) framework with linear utility functions and high-level route characteristics. While several recent studies have started to explore the applicability of deep learning for route choice modeling, they are limited to path-based models with relatively simple model architectures and relying on predefined choice sets. Existing link-based models can capture the dynamic nature of link choices within the trip without the need for choice set generation, but still assume linear relationships and link-additive features. To address these issues, this study proposes a general deep inverse reinforcement learning (IRL) framework for link-based route choice modeling, which is capable of incorporating diverse features (of the state, action and trip context) and capturing complex relationships. Specifically, we adapt an adversarial IRL model to the route choice problem for efficient estimation of context-dependent reward functions without value iteration. Experiment results based on taxi GPS data from Shanghai, China validate the superior prediction performance of the proposed model over conventional DCMs and other imitation learning baselines, even for destinations unseen in the training data. Further analysis show that the model exhibits competitive computational efficiency and reasonable interpretability. The proposed methodology provides a new direction for future development of route choice models. It is general and can be adaptable to other route choice problems across different modes and networks.

The rapid development of information technology has significantly promoted social interactions among people. Social interactions may have become an important channel for commuters to obtain traffic information. Although commuters’ own travel experience and Advanced Traveller Information Systems (ATIS) are acknowledged as two common traffic information sources that can impact commuters’ route choice, the possible influence of social interactions has not yet been empirically confirmed. Therefore, an empirical study is conducted in this paper to explore whether and how social interactions affect commuters’ route choice. The presented study is divided into two phases. First, 1000 commuters are surveyed about their social interactions with other commuters. A total of 809 valid questionnaires are returned from the commuters. The survey confirms that social interactions among the commuters are very common and indeed influence commuters’ route choice. Second, because the essence of the influence of social interactions on commuters’ route choice is an information-adoption process, partial least squares (PLS) component-based structural equation modelling (SEM) is employed to study the key success factors that influence information adoption in commuters’ route choice in the context of social interactions. In the second phase, 236 valid questionnaires were returned. The obtained results demonstrate that information relevance, information accuracy, source expertise, source integrity, usefulness and extraversion are the factors that have significant influence on commuters’ information adoption in the context of social interactions; while the roles of responding, trust and field dependence are not significant. Finally, some discussion and inspirations are provided based on the data analysis results. The findings in the presented study offer some insights into using social interaction information to guide commuters’ route choice.

This paper addresses the suitability of different cartographic design variants for visually communicating recommended routes. We performed a user study, investigating the potential of six different...

The dynamic signal control and route choice equilibrium are usually integrated into a noncooperative game between the network authority and the road users. There are mainly two problems in most existing optimization methods. Firstly, the authority is often placed in the upper level in bi-level programming models, this pure system-optimization-oriented framework may increase the difficulty in obtaining an equilibrium flow distribution. Secondly, the rate of drivers’ compliance on the control strategy has not been fully investigated, which makes the problem intractable in real time, especially in a connected vehicle (CV) environment. This paper proposes a modified Stackelberg games model to change the format of the authority-user and user-authority dynamically. The direct communication between the authority and users is established, and the drivers’ compliance rate is applied as the level-change threshold index. Considering the difference between the drivers’ realized travel time and the predicted travel time on the variable message sign (VMS), a logit model is formed to calibrate the compliance rate in every time step. Based on a modified wavelet neural network algorithm, the model predictive control (MPC) fulfills the level-change procedure using the software Matlab 2018b. Six benchmarks are applied in a numerical example. The results show that the proposed model with the centralized framework obtains the minimum total travel cost compared with the benchmarks. Combined with the real-time mutual feedback between drivers’ response and control strategy, the level-change procedure potentially maintains the compliance rate within a certain level.

The traffic behaviours of commuters may cause traffic congestion during peak hours. Advanced Traffic Information System can provide dynamic information to travellers. Due to the lack of timeliness and comprehensiveness, the provided information cannot satisfy the travellers’ needs. Since the assumptions of traditional route choice model based on Expected Utility Theory conflict with the actual situation, a route choice model based on Game Theory is proposed to provide reliable route choice to commuters in actual situation in this paper. The proposed model treats the alternative routes as game players and utilizes the precision of predicted information and familiarity of traffic condition to build a game. The optimal route can be generated considering Nash Equilibrium by solving the route choice game. Simulations and experimental analysis show that the proposed model can describe the commuters’ routine route choice decisionexactly and the provided route is reliable.

In recent years, the metro system has advanced into an efficient transport system and become the mainstay of urban passenger transport in many mega-cities. Passenger flow is the foundation of making and coordinating operation plans for the metro system, and therefore, a variety of studies were conducted on transit assignment models. Nevertheless route choice sets of passengers also play a paramount role in flow estimation and demand prediction. This paper first discusses the main route constraints of which the train schedule is the most important, that distinguish rail networks from road networks. Then, a two-step approach to generate route choice set in a metro network is proposed. Particularly, the improved approach introduces a route filtering with train operational information based on the conventional method. An initial numerical test shows that the proposed approach gives more reasonable route choice sets for scheduled metro networks, and, consequently, obtains more accurate results from passenger flow assignment. Recommendations for possible opportunities to apply this approach to metro operations are also provided, including its integration into a metro passenger flow assignment and simulation system in practice to help metro authorities provide more precise guidance information for passengers to travel.

© 2017 Elsevier Ltd Explicitly including the dynamics of users’ route choice behaviour in optimal traffic control applications has been of interest for researchers in the last five decades. This has been recognized as a very challenging problem, due to the added layer of complexity and the considerable non-convexity of the resulting problem, even when dealing with simple static assignment and analytical link cost functions. In this work we establish a direct behavioural connection between the different shapes and structures emerging in the solution space of such problems and the underlying route choice behaviour. We specifically investigate how changes in the active equilibrium route set exert direct influence on the solution space's structure and behaviour. Based on this result, we then formulate and validate a constrained version of the original problem, yielding desirable properties in terms of solution space regularity. ispartof: TRANSPORTATION RESEARCH PART B-METHODOLOGICAL vol:117 pages:892-906 ispartof: location:NW Univ, Evanston Campus, Chicago, IL status: published

This paper addresses the stochastic user equilibrium (SUE) in the case where the route choice is the multinomial logit model (MNL). Our main finding is that MNL SUE can be formulated and solved as an immediate extension of the deterministic user equilibrium (DUE) through a particular application of Wardrop’s first principle. The latter states, in general, that at equilibrium, the cost of all used routes is equal and not higher than those of unused routes. The extension is achieved by applying this statement to the “perceived cost” of a choice alternative, which is defined here as its generalized cost plus the logarithm of its choice probability multiplied by the logit parameter. Thus, substituting in DUE models the generalized costs with the perceived costs allows to easily adapt to MNL SUE the existing formulations and algorithms for DUE, as well as to manage a smooth transition of the route choice model from stochastic to deterministic by reducing the logit parameter down to zero. Particular consideration is devoted to the interpretation of the numerical solution as a restricted logit model, where only sufficiently good alternatives receive a positive probability. A family of MNL SUE models is then presented ranging from nonlinear optimization to variational inequalities and fixed-point problems, with both explicit and implicit path enumeration. A range of numerical tests is presented with the aim of assessing the continuity of the model results for decreasing logit parameter and proving the applicability of the proposed approach to real size networks, with particular emphasis on the performance and convergence of the methods.

We propose a route choice model that relaxes the independence from irrelevant alternatives property of the logit model by allowing scale parameters to be link specific. Similar to the recursive logit (RL) model proposed by Fosgerau et al. (2013), the choice of path is modeled as a sequence of link choices and the model does not require any sampling of choice sets. Furthermore, the model can be consistently estimated and efficiently used for prediction. A key challenge lies in the computation of the value functions, i.e. the expected maximum utility from any position in the network to a destination. The value functions are the solution to a system of non-linear equations. We propose an iterative method with dynamic accuracy that allows to efficiently solve these systems. We report estimation results and a cross-validation study for a real network. The results show that the NRL model yields sensible parameter estimates and the fit is significantly better than the RL model. Moreover, the NRL model outperforms the RL model in terms of prediction.

In this paper, we consider captive mode travelers (those who have no other choices but rely on one specific travel mode for daily commuting trips) in a multi-modal network equilibrium (MMNE) problem. Specifically, the dogit model is adopted to account for captive mode travelers in the modal split problem, and the path-size logit (PSL) model is used to capture route overlapping effects in the traffic assignment problem. The dogit-PSL MMNE model is formulated as an equivalent entropy-based mathematical programming (MP) problem, which admits solution existence and uniqueness. Three numerical examples are provided. The first example examines the effects of mode captivity and route overlapping on network performances and observes that accounting for captive mode travelers would produce different equilibrium states and hence the network performance indicators. The second example applies the dogit-PSL MMNE model for evaluating the exclusive bus lane (EBL) expansion plans, in which a consistent synthetic proportional index is proposed. Numerical results show that considering mode captivity may produce substantial impacts on the odds (up to 50 percent of odds in the given scenarios) of making different EBL line expansion decisions. The third example implements the dogit-PSL MMNE model in the Seoul network to show the applicability of the dogit-PSL MMNE model in a real-size multi-modal system.