Traffic state information is crucial for different applications, e.g., in design and operation of road traffic networks, and in navigation services. Traffic sensing data, e.g., loop-detector data, may directly provide the desired information. Alternatively, the traffic state information may be estimated with data that only provides partial and noisy information. To apply this process, i.e., traffic state estimation, we have to make choices related to which data are collected and how these are processed. The macroscopic traffic state can be described using the variables flow, density and mean speed, where flow is equal to the product of density and mean speed. Edie’s generalized definitions of traffic flow define these three variables for spatial-temporal areas. Alternatively, traffic flow can be described using the three dimensions space, time and cumulative flow. The cumulative flow is defined as the cumulative number of vehicles that have passed a position at a specific time, which means that it is a discrete variable. However, the discrete function can be smoothed over space and time. In this case, the macroscopic variables flow and (negative) density can be determined for points in space-time by taking the derivatives to time and space of the smoothed cumulative flow function. In this thesis, a distinction is made between microscopic and macroscopic traffic sensing data. Examples of microscopic traffic sensing data are probe individual speed data and spacing data. Macroscopic data can describe Edie’s generalized definitions of traffic flow for spatial-temporal areas, e.g., probe mean speed data or aggregated double loop-detector data. Alternatively, macroscopic sensing data can describe the change in cumulative flow between points in space-time, e.g., detector count data or relative flow data. The scientific gaps addressed in this thesis are subdivided in four parts that relate to each other. First, we evaluate the errors that are induced when estimating the mean speed for spatial-temporal areas based on error-free data. This provides insight in the errors that arise due to incomplete information and incorrect assumptions when estimating the mean speed. Second, the option to use probe data to mitigate the cumulative count error problem is considered. This problem occurs when estimating the cumulative flow curves based on (stationary) detector data. For this purpose, both probe mean speed and probe trajectory data are used. The probe mean speed data relates to the first part as they describe the mean speed for spatial-temporal areas. If relative flow observations are added to the probe trajectory data, relative flow data from moving observers that are part of the traffic flow are obtained. In the third part, these relative flow data are used to estimate the traffic state. In this part, different combinations of observers are used, which includes stationary observers, moving observers that are part of the traffic flow and moving observers that travel in opposing direction. To estimate the traffic state with relative flow data, streaming-data-driven and model-driven estimation approaches are considered. In a model-driven estimation approach historical data are used to expose traffic flow models. Therefore, we address the possibility to use historical relative flow data to expose these model. The fourth and final part relates the option that road authorities collect personal traffic sensing data (e.g., probe trajectory and/or relative flow data) directly from road-users. In other parts of this thesis, we designed methodologies to use these data, which may be valuable for road authorities. Therefore, it is interesting to investigate how road authorities can gain access to these personal data. @en

Traffic flow can be described using three dimensions, i.e., space x, time t and cumulative flow N. This study considers estimating the cumulative flow over space and time, i.e., N(x,t), using relative flow data collected by stationary and moving observers. Stationary observers, e.g., loop-detectors, can observe flow at fixed position over time. Furthermore, automated or other equipped and connected vehicles can serve as moving observers that observe flow relative to their position over time. To present the value of relative flow data, in this paper, we take the perspective of a model-based estimation approach. In this approach, the data is used in two processes: (1)information assimilation of real-time data and models and (2)learning of the models used in information assimilation based on historical data. This paper focuses on traffic state estimation on links. However, we explain that, in absence of stationary observer that are positioned at the link boundaries, it is valuable to consider the information propagation over nodes. Throughout this study a LWR-model with a triangular fundamental diagram (FD)is used to develop the principles that can be used for the two processes. These principles are tested in a simulation (VISSIM)study. This study shows that we can find the traffic flow model parameters and can partially estimate the link boundary conditions based on relative flow data collected by moving observers alone. It also shows that the traffic flow behavior differs partially from the LWR-model with triangular FD, and therefore, we recommend the option to learn and use other traffic flow models in future research. Overall, relative flow data is considered valuable to obtain model learning datasets and to estimate the traffic state.

@en

Detector data can be used to construct cumulative flow curves, which in turn can be used to estimate the traffic state. However, this approach is subject to the cumulative error problem. Multiple studies propose to mitigate the cumulative error problem using probe trajectory data. These studies often assume “no overtaking” and thus that the cumulative flow is zero over probe trajectories. However, in multi-lane traffic this assumption is often violated. Therefore, we present an approach to estimate the change in cumulative flow along probe trajectories between detectors based on disaggregated detector data. The approach is tested with empirical data and in microsimulation. This shows that the approach is a clear improvement over assuming “no overtaking” in free-flow conditions. However, the benefits are not clear in varying traffic conditions. The approach can be applied in practice to mitigate the cumulative error problem and estimate the traffic state based on the resulting cumulative flow curves. As the performance of the approach depends on the changes in traffic conditions, it is suggested to use the probe speed observations between detectors to assign an uncertainty to the change in cumulative flow estimates. Furthermore, a potential option for future work is to use more elaborate schemes to estimate the probe relative flow between detectors, which may, for instance, combine probe speeds with estimates of the macroscopic states along the probe trajectory. If these macroscopic estimates are based on the cumulative flow curves at the detector locations, this would result in an iterative approach.@en

Demand exceeding the capacity of a bottleneck will create congestion upstream of that bottleneck. Once this congestion occurs, the maximum flow through this bottleneck decreases (capacity drop). By limiting the flow towards the bottleneck, one can prevent or postpone the capacity drop and the accompanying congestion. In case the bottleneck is caused by an on-ramp, a common approach is to meter the on-ramp flow. For metering to be effective the algorithm has to be tuned carefully. Normally, the parameters of a metering algorithm are fit for the situation. However, traffic is dynamic and external factors might change, which both lead to changes in parameters of the traffic process. This paper studies how these parameters can be updated dynamically in the control algorithm. It considers various ramp metering algorithms and introduces methods to adapt their parameters. They are tested with simulations using the METANET model. This shows that parameter adaptation improves traffic state. Gains in travel time due to parameter adaptation are typically several percent compared to non-adaptive ramp metering. Road authorities can use these findings to improve ramp metering algorithms and reduce delays @en

The Macroscopic Fundamental Diagram (MFD) describes the relation between accumulation and speed in a zone. While theoretically expected, empirical validations have been done with limited numbers of floating car (e.g., taxi) data, and loop detectors. This paper will verify existence, shape and crispness of the MFD using Floating Car Data (FCD) from Google. Due to the large amount of users (i.e., high penetration rate), this unique data-set contains traffic information with a high spatial resolution, a high spatial scope and high reliability. We use the data for 3 purposes. First, an MFD for the city of Amsterdam is constructed, revealing a strong relationship between the average density and the average flow. It also shows that the urban road network never reaches its capacity. Secondly, inhomogeneity is analysed. Traffic is well spread over the network, hence inhomogeneity is low. Moreover, if present, the inhomogeneity has only a minor effect on the flow. Also traffic in different directions is homogeneous. Thirdly, for the first time, an MFD is created for a whole country, which turned out to be very crisp as well. This suggests that small areas or a directional split are not needed to create crisp MFDs. That, in turn, implies that a crisp MFD is not a sufficient condition to apply control without considering internal dynamics.
@en

This article presents a procedure to estimate the macroscopic traffic state in a pre-defined space-time mesh using relative flow data collected by stationary and moving observers. The procedure consist of two consecutive and independent processes: (1) estimate point observations of the cumulative vehicle number in space-time, i.e., N(x, t), based on relative flow data from the observers and (2) estimate flow and density in a pre-define space-time mesh based on the point observations of N. In this paper, the principles behind the first process are explained and a methodology (the Point-Observations N (PON) estimation methodology) is introduced for the second process. This methodology does not incorporate information in the form of a traffic flow model or historical data. To evaluate this performance and improve our understanding of the methodology, a microscopic simulation study is conducted. The estimation performance is effected by the homogeneity and stationarity of traffic in estimation area and in the sample area. In case of large changes in traffic conditions, e.g., from free-flow to congestion or stop-and-go waves, a better sampling resolution will improve localizing these changes in space and time and hence improve the estimation performance. In the simulation study, the proposed methodology is also compared with estimates based on loop-detector data. This indicates that the combination of the proposed methodology and data yields an alternative for existing combinations of methodology and data. Especially, in terms of density estimation the introduced methodology shows promising results.

@en

In ruwe vorm zijn floating car data rijk en uiterst gedetailleerd. Het punt is alleen dat we om privacyredenen niet met die ruwe data mogen werken – we moeten het doen met geaggregeerde data die door een privacyfilter zijn gehaald. Is daarmee nog enige verkeersmanagement-eer te behalen? Jawel, mits je de FCD slim combineert met bijvoorbeeld lusdata. Zelfs het real-time schatten van wachtrijen behoort dan tot de mogelijkheden, aldus de auteurs.@en

The fundamental diagram (FD) describes the relation between the flow and density in equilibrium conditions. In this paper, we propose an estimation approach to estimate the FD based on data from moving observers. This approach consists of two main steps: (1) estimate flow and density for space-time areas based on trajectories of moving observers and the times and locations they are overtaken or being overtaken and (2) estimate the FD based on the fflow,densityg-estimates. To evaluate and gain a deeper understanding of the proposed approach, a simulation study was conducted. This study shows that the fflow,densityg-estimates provide valuable information to estimate the FD. Furthermore, the FDs belonging to the simulated traffic flow are estimated accurately. We realize that the second step is expected to be less accurate for traffic that behaves stochastic. Therefore, we provide a potential solution path to extend the second step in future work. @en

Traffic state estimation is a crucial element in traffic management systems and in providing traffic information to road users. In this article, we evaluate traffic sensing data-based estimation error characteristics in macroscopic traffic state estimation. We consider two types of sensing data, that is, loop-detector data and probe speed data. These data are used to estimate the mean speed in a discrete space-time mesh. We assume that there are no errors in the sensing data. This allows us to study the errors resulting from the differences in characteristics between the sensing data and desired estimate together with the incomplete description of the relation between the two. The aim of the study is to evaluate the dependency of this estimation error on the traffic conditions and sensing data characteristics. For this purpose, we use microscopic traffic simulation, where we compare the estimates with the ground truth using Edie's definitions. The study exposes a relation between the error distribution characteristics and traffic conditions. Furthermore, we find that it is important to account for the correlation between individual probe data-based estimation errors. Knowledge related to these estimation errors contributes to making better use of the available sensing data in traffic state estimation.

@en