P.H.L. Bovy

Pedestrian route-choice and activity scheduling theory and models

Among the most interesting and challenging theoretical and practical problems in describing pedestrians behavior are route choice and activity scheduling. Compared to other modes of transport, a characteristic feature of pedestrian route choice is that routes are continuous trajectories in time and space: since a pedestrian chooses a route from an infinite set of alternatives, dedicated theories and models describing pedestrian route choice are required. This article puts forward a new theory of pedestrian behavior under uncertainty based on the concept of utility maximization. The main behavioral assumption is that pedestrians optimize some predicted pedestrian-specific utility function, representing a trade-off between the utility gained from performing activities at a specific location, and the predicted cost of walking subject to the physical limitations of the pedestrians and the kinematics of the pedestrian. The uncertainty reflects the randomness of the experienced traffic conditions. Based on this normative theory, route choice, activity area choice, and activity scheduling are simultaneously optimized using dynamic programming for different traffic conditions and uncertainty levels. Throughout the article, the concepts are illustrated by examples.

State-of-the-art of vehicular traffic flow modelling

Abstract Nowadays traffic flow and congestion is one of the main societal and economical problems related to transportation in industrialized countries. In this respect, managing traffic in congested networks requires a clear understanding of traffic flow operations; i.e. insights into what causes congestion, what determines the time and location of traffic breakdown, how does the congestion propagate through the network, etc., are essential. For this purpose, during the past fifty years, a wide range of traffic flow theories and models have been developed to answer these research questions. This paper presents a overview of some fifty years of modelling vehicular traffic flow. A rich variety of modelling approaches developed so far and in use today will be discussed and compared. The considered models are classified based on the level-of-detail with which the vehicular flow is described. For each of the categories, issues like modelling accuracy, applicability, generalizability, and model calibration and validation are discussed.

Extended time-to-collision measures for road traffic safety assessment.

This article describes two new safety indicators based on the time-to-collision notion suitable for comparative road traffic safety analyses. Such safety indicators can be applied in the comparison of a do-nothing case with an adapted situation, e.g. the introduction of intelligent driver support systems. In contrast to the classical time-to-collision value, measured at a cross section, the improved safety indicators use vehicle trajectories collected over a specific time horizon for a certain roadway segment to calculate the overall safety indicator value. Vehicle-specific indicator values as well as safety-critical probabilities can easily be determined from the developed safety measures. Application of the derived safety indicators is demonstrated for the assessment of the potential safety impacts of driver support systems from which it appears that some Autonomous Intelligent Cruise Control (AICC) designs are more safety-critical than the reference case without these systems. It is suggested that the indicator threshold value to be applied in the safety assessment has to be adapted when advanced AICC-systems with safe characteristics are introduced.

Continuum modeling of multiclass traffic flow

In contemporary macroscopic traffic flow modeling, a distinction between user-classes is rarely made. However, it is envisaged that both the accuracy and the explanatory ability of macroscopic traffic flow models can be improved significantly by distinguishing classes and their specific driving characteristics. In this article, we derive such a multiple user-class traffic flow model. Starting point for the derivation of the macroscopic flow model is the user-class specific phase-space density, which can be considered as a generalization of the traditional density. The gas-kinetic equations describing the dynamics of the multiclass Phase-Space Density (MUC-PSD) are governed by various, interacting processes, such as acceleration towards a class-specific desired velocity, deceleration caused by vehicle interactions and the influence of lane changing. The gas-kinetic equations serve as the foundation of the proposed macroscopic traffic flow models, describing the dynamics of the class-dependent spatial density, velocity and velocity variance. The modeling approach yields explicit relations for both the velocity and the velocity variance. These equilibrium relations show competing processes: on the one hand, drivers accelerate towards their class-dependent desired velocity, while on the other hand, they need to decelerate due to interactions with vehicles from their own class and asymmetric interactions with vehicles from other classes. Using the operationalized model, macroscopic simulations provide insight into the model behavior for different scenarios.

ASSESSMENT OF ROADWAY CAPACITY ESTIMATION METHODS

Capacity is a central concept in roadway design and traffic control. Estimation of empirical capacity values in practical circumstances is not a trivial problem; it is very difficult to define capacity in an unambiguous manner. Empirical capacity estimation for uninterrupted roadway sections has been studied. Headways, traffic volumes, speed, and density are traffic data types used to identify four groups of capacity estimation methods. Aspects such as data requirement, location choice, and observation period were investigated for each method. The principles of the different methods and the mathematical derivation of roadway capacity are studied and discussed. Among the methods studied are the headway distribution approaches, the bimodal distribution method, the selected maxima, and the direct probability method. Of the methods based on traffic volume counts, the product limit method is recommended for practical application because of sound underlying theory. An example of the application of this promising method is presented. Attempts to determine the validity of existing roadway capacity estimation methods were disappointing because of the many ambiguities related to the derived capacity values and distributions. A reliable and meaningful estimation of capacity is not yet possible. Lack of a clear definition of the notion of capacity is the main hindrance in understanding what exactly represents the estimated capacity value or distribution in the various methods. If this deficiency is corrected, promising methods for practical use in traffic engineering are the product limit method, the empirical distribution method, and the well-known fundamental diagram method, in that order. The choice of a particular method strongly depends on the available data.

Gas-Kinetic Modeling and Simulation of Pedestrian Flows

Insight into pedestrian flow operations is important in both planning and geometric design of infrastructure facilities such as railway stations as well as in the management of pedestrian flows in such facilities. Lack of empirical knowledge regarding the characteristics of pedestrian flows under varying circumstances and designs motivates using a model-based approach. In this study, a new pedestrian flow model based on the gaskinetic modeling paradigm is established. The mesoscopic equations describe the dynamics of so-called pedestrian phase-space density, which can be considered as a two-dimensional generalization of the phase-space density used in gas-kinetic vehicular traffic flow. Convection, acceleration, and noncontinuum transition terms govern the dynamics. The latter terms reflect the dynamic influence of pedestrians decelerating and the changing angle of movement due to pedestrians interacting. Numerical solutions of the resulting gas-kinetic equations are established by using a novel particle discretization approach. Essentially, this approach upgrades the mesoscopic equations to a microscopic pedestrian flow simulation model. Using the particle discretization approach, the model’s behavior is tested for different test-case scenarios. The model is shown to produce plausible speed-density functions from which walking speeds and travel times can be derived for a variety of conditions.

On Modelling Route Choice Sets in Transportation Networks: A Synthesis

Abstract This article reviews a number of topics related to the modelling and generation of route choice sets, specifically for applications in large networks. It synthesizes existing knowledge using a conceptual framework, and addresses in what respects route choice differ from other travel choices. It shows that it is advantageous to distinguish the processes of choice set formation and choice per se, but also to explicitly separate the modelling steps of choice set generation and choice modelling. The article discusses the various purposes for which route choice sets may be used and what these mean for practical choice set modelling. A generic conceptual scheme is presented relating the distinct key elements of the known route choice set generation approaches aimed at their classification and characterization. Some indications for their empirical validity are presented derived from applications to various uni‐modal and multi‐modal networks.

Quasi-variational inequality formulation of the multiclass dynamic traffic assignment problem ☆

We consider the extension of a single user-class macroscopic dynamic traffic assignment model to include multiple user-classes. The distinction between user-classes is typically based on vehicle characteristics, e.g. cars and trucks. Interactions between the user-classes sharing the same road infrastructure are taken into account. To deal with various different asymmetries that may occur, such as interuser-class interaction, interspatial and intertemporal asymmetries, the model is specified as a (quasi) variational inequality problem. A nested modified projection method is proposed to solve the assignment problem. The solution of the problem depends heavily on the choice of some very important input: the multiclass link travel time functions. Under mild restrictions there exists a solution, which needs however not be unique. A case study illustrates the model.

Analysis of attrition biases and trip reporting errors for panel data

Abstract The objective of the study is to assess the nature and magnitude of reporting errors and attrition biases in the first two waves of a household panel survey data set that consists of weekly trip diaries. This paper investigates the relationship between reporting errors in a first-wave diary and the subsequent decision to participate in a second-wave survey and the relationship between the same first-wave errors and the errors in the second-wave survey for those households which chose to participate. An econometric procedure is developed which accounts for attrition bias and relates reporting errors from the two waves. The model system offers a mechanism by which relationships among mobility, trip reporting errors, and attrition behavior in a panel survey can be statistically evaluated. It consists of three elements: wave-one trip equation, attrition probability model, and wave-two trip equation. The conditional distribution of the error terms of these three components is evaluated assuming chronological dependency among the errors. The result is used to develop a correction term for attrition bias and a term which accounts for the dependency between the wave-one error and wave-two error. Results from empirical application of the model system are reported together with the use of the attrition model to develop an efficient sample weighting factor that fully utilizes the information in the survey results.

Stochastic Route Choice Set Generation: Behavioral and Probabilistic Foundations

This article addresses the generation of choice sets for the purposes of route choice analysis and flow prediction in uni-modal and multi-modal networks. Ample attention is devoted to the implications for choice set generation of these various purposes. Based on a theory on choice behavior, a new model-based choice set generation approach will be elaborated, called the doubly stochastic choice set generation model, meant for establishing choice sets prior to the choice modeling step. Because of its stochastic principle, a typical property of the proposed generation approach is that the size and composition of the generated choice sets are stochastic variables. We will devote ample attention to these stochastic properties using theoretical derivations and experimental studies. The article reports on the calibration of this generation approach for multi-modal networks and illustrates the approach with a number of predictions in various networks.