Gunnar Flötteröd

Bayesian Demand Calibration for Dynamic Traffic Simulations

We present an operational framework for the calibration of demand models for dynamic traffic simulations, where calibration refers to the estimation of a structurally predefined models parameters from real data. Our focus is on disaggregate simulators that represent every traveler individually. We calibrate, also at an individual level, arbitrary choice dimensions within a Bayesian framework, where the analysts prior knowledge is represented by the dynamic traffic simulator itself and the measurements are comprised of time-dependent traffic counts. The approach is equally applicable to an equilibrium-based planning model and to a telematics model of spontaneous and imperfectly informed drivers. It is based on consistent mathematical arguments, yet it is applicable in a purely simulation-based environment and, as our experimental results show, is capable of handling large scenarios.

Estimating network properties from snowball sampled data

This article addresses the estimation of topological network parameters from data obtained with a snowball sampling design. An approximate expression for the probability of a vertex to be included in the sample is derived. Based on this sampling distribution, estimators for the mean degree, the degree correlation, and the clustering coefficient are proposed. The performance of these estimators and their sensitivity with respect to the response rate are validated through Monte Carlo simulations on several test networks. Our approach has no complex computational requirements and is straightforward to apply to real-world survey data. In a snowball sample design, each respondent is typically enquired only once. Different from the widely used estimator for Respondent-Driven Sampling (RDS), which assumes sampling with replacement, the proposed approach relies on sampling without replacement and is thus also applicable for large sample fractions. From the simulation experiments, we conclude that the estimation quality decreases with increasing variance of the network degree distribution. Yet, if the degree distribution is not to broad, our approach results in good estimates for the mean degree and the clustering coefficient, which, moreover, are almost independent from the response rate. The estimates for the degree correlation are of moderated quality.

Some practical extensions to the Cell Transmission Model

This article describes some practical extensions to Daganzos Cell Transmission Model. Flow calculations for straight, merge, and diverge cells are subsumed in a single computation scheme which allows for arbitrary cell connec- tivity. Since it is planned to apply the resulting model in a mathematical programming based traffic monitoring system, approximate sensitivities are also provided. I. INTRODUCTION Any intelligent transportation systems performance sig- nificantly depends on its capability to reproduce and predict the state of the traffic system under consideration. From a control engineering point of view, this can be achieved by representing the system in state space form and then applying an observation algorithm such as Kalman filtering (1), (10). Pursuing this idea, we currently are in the process of setting up a traffic model with certain properties we found most appropriate in a state estimation context. Specifically, this article describes a network loading model, which is (like all other components of the overall system representation) a deterministic macroscopic discrete time model that can at least approximately be differentiated. Although we sacrifice many benefits of microscopic ap- proaches by choosing an aggregated model (2), (8), we still expect meaningful estimation results due to the straight- forward applicability of a large number of mathematical programming algorithms to our problem formulation (as compared to the far more difficult handling of microscopic traffic simulator outputs) (3), (9). Since we also work on agent based models, this work can be understood to comple- ment such approaches. To build on a strong foundation, we chose the Cell Transmission Model (CTM) as a starting point. It fulfills the aforementioned requirements and has proven to be a proper numerical approximation of the well-established LWR representation of traffic flow (4), (7). Although it was originally proposed for highway traffic, we intend to apply it also to urban systems, where traffic intrinsic dynamics become secondary compared to the effect of light signals. In this case, even simpler models appear to be sufficient to capture relevant effects (6). Still, the possibility for seamless transition from freeway to urban traffic using a single model makes the CTM even more attractive. In addition, it is a first order model with an accordingly low number of states and parameters to be estimated.

Capturing Dependency Among Link Boundaries in a Stochastic Dynamic Network Loading Model

This work adds realistic dependency structure to a previously developed analytical stochastic network loading model. The model is a stochastic formulation of the link-transmission model, which is an operational instance of Newells simplified theory of kinematic waves. Stochasticity is captured in the source terms, the flows, and, consequently, in the cumulative flows. The previous approach captured dependency between the upstream and downstream boundary conditions within a link i.e., the respective cumulative flows only in terms of time-dependent expectations without capturing higher-order dependency. The model proposed in this paper adds an approximation of full distributional stochastic dependency to the link model. The model is validated versus stochastic microsimulation in both stationary and transient regimes. The experiments reveal that the proposed model provides a very accurate approximation of the stochastic dependency between the links upstream and downstream boundary conditions. The model also yields detailed and accurate link state probability distributions.

Analysis of Implicit Choice Set Generation Using a Constrained Multinomial Logit Model

Discrete choice models are defined conditional to the analysts knowledge of the actual choice set. The common practice for many years has been to assume that individual-based choice sets can be deterministically generated on the basis of the choice context and characteristics of the decision maker. This assumption is not valid or not applicable in many situations, and probabilistic choice set formation procedures must be considered. The constrained multinomial logit model (CMNL) has recently been proposed as a convenient way to deal with this issue, as it is also appropriate for models with a large choice set. In this paper, how well the implicit choice set generation of the CMNL approximates the explicit choice set generation is analyzed as described in earlier research. The results based on synthetic data show that the implicit choice set generation model may be a poor approximation of the explicit model.

Enhancing MATSim with capabilities of within-day re-planning

This paper presents a framework for simulation of within-day re-planning for the MATSim project. Three major building blocks are presented, each of which represents specific aspects of driver behavior. These components comprise (i) the provision of descriptive information in the form of link travel costs, (ii) prescriptive information in the form of routes, and (iii) a model of driver satisfaction. An exemplary model is presented, which focuses on en-route re-planning under different types of information provision. In this model driver perception is constrained to link traversal costs and decisions are made by application of a standard shortest path algorithm. The satisfaction of a traveler is modeled with a scoring (utility) function that evaluates routes as well as activities travelers are aiming at. The frameworks applicability is tested with a simple fictive network and a real-world network of Greater Berlin.

Towards system optimum: finding optimal routing strategies in time-dependent networks for large-scale evacuation problems

Evacuation planning crucially depends on good routing strategies. This article compares two different routing strategies in a multi-agent simulation of a large real-world evacuation scenario. The first approach approximates a Nash equilibrium, where every evacuee adopts an individually optimal routing strategy regardless of what this solution imposes on others. The second approach approximately minimizes the total travel time in the system, which requires to enforce cooperative behavior of the evacuees. Both approaches are analyzed in terms of the global evacuation dynamics and on a detailed geographic level.

High Speed Combined Micro/Macro Simulation of Traffic Flow

We describe two new and practically relevant simulation techniques related to the kinematic wave model of traffic flow. Firstly, we demonstrate how the well-known Godunov solution scheme can be run on variable time scales in a computationally very efficient way. Secondly, we demonstrate how the resulting macroscopic traffic flow model can be run in conjunction with a microscopic model of driver behavior while maintaining high computational performance.

A CA model for bidirectional pedestrian streams

The modeling of pedestrian flows is of particular importance for the planning of pedestrian facilities and the preparation for emergency situations. Even though there is a wide variety of simulation models on the market, a simulation model that deals with bidirectional pedestrian flow adequately is still missing. This contribution proposes an event-based cellular automaton model that is capable of simulating bi- and unidirectional pedestrian streams. The model is built on a theoretically sound foundation. Its performance is demonstrated by a comparison to empirical data.

Disaggregate Path Flow Estimation in an Iterated Dynamic Traffic Assignment Microsimulation

This article describes the first application of a novel path flow and origin/destination (OD) matrix estimator for iterated dynamic traffic assignment (DTA) microsimulations. The presented approach, which operates on a trip-based demand representation, is derived from an agent-based DTA calibration methodology that relies on an activity-based demand model (Flötteröd, Bierlaire, & Nagel, 2011). The objective of this work is to demonstrate the transferability of the agent-based approach to the more widely used OD matrix-based demand representation. The calibration (i) operates at the same disaggregate level as the microsimulation and (ii) has drastic computational advantages over conventional OD matrix estimators in that the demand adjustments are conducted within the iterative loop of the DTA microsimulation, which results in a running time of the calibration that is in the same order of magnitude as a plain simulation. We describe an application of this methodology to the trip-based DRACULA microsimulation and present an illustrative example that clarifies its capabilities.