Search

Although there exists algorithms that give speed advice for cyclists when approaching traffic lights with uncertainty in the timing, they all need to know, and thus assume, the cyclist’s response to the advice in order to be able to optimize the advice. To relax this assumption, in this paper an algorithm is proposed that combines reinforcement learning and planning to learn the reaction of cyclist to the advice and deploys this information for planning the best next advice on-the-fly. Rather than a single search procedure, which is conventional in the existing architectures, two sample-based search procedures are suggested to be used in the algorithm. This makes it possible to obtain an accurate local approximation of the action-value function, in spite of the short computation time that is available in each decision epoch. The algorithm is tested in a simulation case study where the impact of a proper initialisation of action-value function as well as the importance of using two search procedures are affirmed.

Pollution and congestion are important issues in urban mobility. These can potentially be solved by multimodal transport, such as the bicycle-Transit combination, which benefits from the flexible aspect of the bicycle and the wider spatial range of public transport. In addition, the bicycle can increase the catchment areas of public transport stops. Most transit operators consider a fixed 400m buffer catchment area. Currently, not much is known about what influences the size of catchment areas, especially for the bicycle as a feeder mode. Bicycles allow for reaching a further stop in order to avoid a transfer, but it is not clear whether travelers actually do this. This paper aims to fill this knowledge gap by assessing which factors affect feeder distance and feeder mode choice. Data are collected by an on-board transit revealed preference survey among tram travelers in The Hague, The Netherlands. Both regression models and a qualitative analysis are performed to identify the factors that influence feeder distance and feeder mode choice. Results show that the median walking feeder distance is 380m, and the median cycling feeder distance is 1025m. The tram stop density and chosen feeder mode are most important in feeder distance. For feeder mode choice, the following factors are found to be influential: Tram stop density, availability of a bicycle, and frequency of cycling of the tram passenger. Furthermore, the motives of respondents for choosing a stop further away are mostly related to the quality of the transit service and comfort matters, of which avoiding a transfer is named most often. In contrast, the motives for cycling relate mostly to travel time reduction and the built environment. Three important barriers for the bicycle-Tram combination have been discovered: unavailability of a bicycle, insufficient and unsafe bicycle parking places. Infrequent users of the bicycle-Tram combination are more inclined to travel further to a stop that suits them better.

Vehicle bunching often occurs in high-frequency transit systems leading to deterioration of service reliability. It is thus necessary to control vehicles during operations. Holding control is a common solution for this situation, but it may result in longer vehicle running times. Speed adjustments can contribute to more regular operations while preventing prolonged trip times. This paper proposes a control strategy, which combines these two strategies to maintain the regularity of transit operations. The findings based on simulation study for trunk bus services in the Netherlands suggest that combining the two strategies implies both the positive and negative attributes of each control.

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.

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 {flow, density}-estimates. To evaluate and gain a deeper understanding of the proposed approach, a simulation study was conducted. This study shows that the {flow, density}-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.

Demand for (public) transportation is subject to dynamics affected by technological, spatial, societal and demographic aspects. The political environment, together with financial and spatial constraints limit the possibilities to address transit issues arising from growing demand through the construction of new infrastructure. Upgrading of existing services and improving integration over the entire trip chain are two options that can address these transport issues. However, transport planners and transport service operators often fail to include the entire trip when improving services, as improvement is normally achieved through the adaptations of characteristics (e.g. speeds, stop distances) of the services. Our developed framework consists of two parts: one to assess the characteristics of the different bus services and their access and egress modes, and one to assess the effects of integration of these services, which includes the modelling and analysis in a regional transit model. The framework has successfully been applied to a case study showing that bus systems with higher frequencies and speeds can attract twice the amount of cyclists on the access and egress sides. It also shows that passengers accept longer access and egress distances with more positive characteristics of the bus service (higher speeds, higher frequencies).

Summarization: Many network-wide traffic management strategies based on the Macroscopic or Network Fundamental Diagram (NFD) have been developed in recent years. In this paper we investigate one of these strategies, namely feedback NFD-based perimeter control, which limits the rate at which vehicles are allowed to enter an urban region. The inflow limitation is imposed at selected main entry links (gated links) and results in queuing of vehicles at the boundaries of the network. Most of the works on this subject neglected the effect of the queued vehicles on the traffic conditions upstream of the gated links. In this paper we analyze in microsimulation the queuing at the gated links and its effect on network performance for a realistic network model. Moreover, a queue management strategy based on a continuous quadratic knapsack problem aiming at balancing relative queues at the gated links is proposed. Simulation results suggest that queues under perimeter control are shorter in space and time than with no perimeter control. Additionally, managing the queues at the gated links not only improves the overall network performance but also reduces the possibility of queue propagation to the upstream junctions. This improves traffic flow outside the protected network. Παρουσιάστηκε στο: 19th IEEE International Conference on Intelligent Transportation Systems

Traffic signals are not only useful facilities to ensure the safety at an intersection, but also an important traffic management tool to improve the urban traffic network performance. Many traffic signal strategies have been presented to increase the throughput of an urban network and reduce the total delay. The difficulty is that a successful local optimization does not mean a better global performance and that the high complexity makes centralized coordinated strategies usually not efficient for a large network. To overcome these drawbacks, some promising distributed strategies are presented, such as the back-pressure algorithm. Whereas the back-pressure algorithm overcomes the main fundamental problems, some practical issues are not considered. Firstly, the original back-pressure algorithm does not take the all red time into consideration. Secondly, the algorithm relies strongly on the loop detector, which decreases the robustness. Therefore, this paper shows a dynamic green time approach which overcomes these drawbacks of the back-pressure algorithm. In the approach, the green time length depends on two elements: the back-pressure at the intersection and the upstream queue length. Meanwhile the green time for each phase is restricted to durations between 15 seconds and 65 seconds to ensure the robustness. The method is tested in a simulation. This shows that optimal dynamic green time approach shows the best performance among other green time mechanisms, such as fixed green time approach. The green time approach not only makes the back-pressure more practical, but also keeps the good network performance of the original algorithm.

Coordinated ramp metering is an effective means to reduce freeway bottleneck delay by postponing the moment of a flow breakdown and the associated capacity drop. To this aim, vehicles are stored at upstream located ramps to reduce the flow towards the bottleneck, causing delays at the ramps to both vehicles that move towards the bottleneck and to those that leave the network elsewhere. There are several different network characteristics that determine the resulting total ramp and bottleneck delays, and hence, whether it is beneficial to include a ramp into the set of coordinated ramps. In this contribution an approach is put forward to make an adequate decision on which ramps should be coordinated given the situation at hand. To quantify the resulting ramp and bottleneck delays, cumulative inflow and outflow curves are developed as a function of the involved situation-specific variables. This enables us to determine the optimal set of coordinated ramps beforehand and to gain insight into the effects of the system variables that determine the delays. By means of a test case the effects are illustrated and a worked example is given of finding such optimal set of coordinated ramps. Results show that the duration of the peak period and the size of the capacity drop strongly determine the delay at the bottleneck, and hence the coordination benefits of postponing the moment of the capacity drop. These variables also have a large influence on the minimum required share of ramp traffic to the bottleneck (i.e. break even share), that determines whether it is beneficial to include the ramp into the coordination.

Traveler behaviors during emergency evacuation are driven by various factors that are potentially different from those under normal traffic conditions. First, a traveler tends to concern more with the prevailing situation than with long-term interests. Second, both traveler's psychological behaviors and route choice decisions are highly related to the availability and quality of traffic information. Third, evacuation instructions issued by management authorities may also affect the decisions of traveler's route choice. In this paper, evacuation route choice behavior is formulated as the integration of perception behavior on route cost and compliance behavior with the instructed route, which can be taken as the generalized model of the classical logit model. A fuzzy inference system (FIS) is adopted to capture the uncertainty and variability of traveler's perception on route cost by adding a dynamic weight to each link, which takes the level of traffic information and distance from traveler to link as two inputs. Macroscopic simulation framework is proposed to validate the evacuation behavior model using FIS. The case study shows that evacuation efficiency is overestimated by classical logit model, and relative low level and high frequency update of traffic information is recommended. In addition, strong enforce-ment of traffic management contributes to high evacuation efficiency but fully mandatory management is unnecessary.

Sags are segments of the road where there is a significant change in gradient from downhill to uphill in a short distance. Empirically, it has been observed that drivers do not compensate adequately for the changing grade resistance force at sags, which limits vehicle acceleration. As a consequence, congestion forms at sags. This paper proposes and compares the implementation of Adaptive Cruise Control (ACC), Traffic State-Adaptive ACC (TSA-ACC), and Cooperative ACC (CACC) at sags. The controllers are implemented in a traffic simulation environment with a single-lane road stretch and a platoon of 200 vehicles of which 10% follow the movements set by the controller and 90% follow the normal driving rules. The simulation results show that all controllers reduce the travel time of the equipped vehicles and have significantly positive effects on the total travel time of the platoon. TSA-ACC produces the most travel time savings. The findings of this paper show that the proposed controllers can be used to improve the performance of traffic flow at sags.

Literature indicates that on a freeway the capacity is higher than the queue discharge rate. This is called the capacity drop. Recent studies show that the queue discharge rate increases as the speed in congestion increases. For many applications, such as dynamic speed limit control design and evaluation, macroscopic models are needed. A kinematic wave model aiming to assess consequences of control strategies should firstly reproduce different queue discharge rates depending on congestion states, and secondly capture the capacity drop for both standing queues and stop-and-go waves. However, no such kinematic wave model exists until now. Therefore, this paper fills in the gap, incorporating the capacity drop into a kinematic wave model. The new model can reproduce different queue discharge rates corresponding to the vehicular speed in congestion, which means the new model can reproduce speed limit effects accurately. By formulating the model in Lagrangian coordinates and applying an upwind discretization method, this paper presents a hybrid model which captures the capacity drop with stop-and-go waves and can be applied to reproduce the capacity drop with standing queues. A case study shows the working of the model with two different queues.

Capacity drop is a common phenomenon that typically occurs at the downstream of freeway congestion area, which decreases traffic operation efficiency significantly. It can be avoided by some traffic control measures such as ramp metering. In this paper, a generic first-order traffic flow model is proposed, to reproduce the capacity drop at both on-ramp bottleneck and lane drop bottleneck. Based on this model, a linear quadratic model predictive control strategy for the integration of dynamic route guidance and ramp metering is presented. The objective is to minimize the total time spend of a traffic network. Due to the linearity of the optimization problem, the problem can be solved efficiently, which makes the problem suitable for real-time traffic control. A hypothetical traffic network containing both on-ramp bottleneck and lane drop bottleneck is utilized as a test bed, to demonstrate the effectiveness of the proposed control strategy.

In the recent past, numerous Variable Speed Limit (VSL) approaches have been designed to resolve traffic jams on the freeway. In this work, the theory of one such VSL strategy -- COoperative Speed Control ALgorithm (COSCAL v2) -- is analytically extended for integration with Ramp Metering (RM), based on traffic flow theory. COSCAL v2 is a reactive feedback VSL strategy that resolves jam waves using shockwave theory. However, on-ramp flows (with or without RM) are currently not included in its design. When VSLs are activated by jams near an on-ramp, the additional merging flow can reduce effectiveness of the VSL scheme, by preventing jam resolution or even triggering new jams in the controlled section. The integrated approach improves effectiveness of the VSL strategy, and overall network efficiency. The extended theory is evaluated for a freeway section with an on-ramp in micro-simulation, and results indicate a lower total time spent as compared to the original algorithm.

This paper proposes an efficient model predictive control strategy that is based on the parameterization of a variable speed limit control scheme. Due to the paramiterization, the solution spaces reduces which leads to improved computation time. The parameterized control scheme consist of a speed-limited area in which a constant speed limit is imposed. By changing the position of the head and tail of this speed-limited area over time it is possible to change the density and flow in and downstream of this area. The controller optimizes the location of the head and tail of this area over time in such a way that the flow into a bottleneck or jam wave is changed such that congestion can be prevented or resolved. An advantage of this approach is that the complexity of the optimization problem does not increase with an increase in the number of variable speed limit gantries. The controller is tested using a second-order macroscopic traffic flow model. It is shown that the controller improves the total time spent by all the vehicles in the network with 3.7% compared to the no-control situation. This improvement is realized by resolving a jam wave. It is also shown that the controller can achieve a better performance then other model predictive control strategies, when using the same amount of computation time.

In this paper, an ex-ante data analysis method is developed for the assessment of the potential of improving freeway throughput by means of variable speed limits. Ex-ante evaluation of dynamic traffic management measures can provide valuable insight into the potential benefits of these measures. This information is needed when planning to implement dynamic traffic management measures. Ex-ante evaluations are usually performed by means of a simulation study. However, it is hard to reproduce realistic traffic flow characteristics in a simulation. This makes it hard to assess the potential benefits of a dynamic traffic management measure. An ex-ante data analysis, however, can provide insight into the benefits of such a measure. Therefore, a theoretical framework will be provided for analyzing the potential of jam prevention at fixed, infrastructural bottlenecks first. Then, a similar framework is developed to assess moving jam resolvability. Finally, this method is applied to off-line inductive loop detector data of the A13 freeway in the Netherlands in order to assess moving jam resolvability. It is found that approximately 18.7% of the moving jams on the A13 freeway are assessed as resolvable with variable speed limits.

In the first phase of the large-scale field trial integrated traffic management in Amsterdam (acronym: PPA), a hierarchical control approach is used to coordinate the algorithms of the ramp metering installations along a densely used freeway corridor with the intersection controllers along one of the connecting urban arterials. This hierarchical control approach requires a similarly organised hierarchical monitoring approach. This approach is based on several freeway and urban state estimation and prediction techniques. In this contribution the authors describe and discuss two of these in terms of the authors underlying rationale and give some some examples of their application. In the final part of this paper the authors discuss a number of critical issues that came up in the implementation of these methods in the PPA. These relate to the difficulty of communicating the underlying scientific ideas into practice. They also relate to the fact that some of the underlying (scientific) puzzles have not yet been solved.

Important statistical characteristics of traffic can be examined via the reconstruction of a so-called phase-space where each point in that space represents a state of traffic and vice versa. Statistical tools such as recurrence analysis can then be used to visualize the temporal evolution of traffic and to assess its non-stationarity structure as an indicator for irregular behaviour. In this paper we investigate, examine and analyze the short-term non-stationary features present in traffic flows based on the historical data available from stop-line detectors collected from a small number of intersections in the Melbourne CBD, Australia and demonstrate that recurrence analysis has considerable potential in identifying changes within the temporal behaviour of traffic.

Around 2 million visitors reach the beach of Copacabana in Rio de Janeiro on the New Years Eve every year. More than 100.000 visitors travel by Metro and around 70% use the Cardeal Arcoverde Station (CAV) as their destination. This creates large pedestrian flows inside the station causing major discomfort and endangering the station users. With the steady increase of the flows in recent years a system engineering approach is applied to mitigate the circulation problems. The approach identified the station (and its components) as an open system with a feedback mechanism (crowd management measures). The system itself is connected to other systems such as the station surroundings and the trains. A system analysis, determined the individual capacities of the station components, their dynamic properties, mutual influences and dependency to the other systems. Based on these, crowd-management measures were devised and enforced in a contingency plan that improved the safety and comfort of pedestrians during the 2012-2013 event. The bottlenecks were better understood and their capacity improved causing significant less crowding in the station and allowing a rapid action of marshals when an emergency occurred. We conclude by showing the improved results that a system engineering approach is a useful approach to plan and access large crowds events.

Multi-scale traffic models aim to describe traffic operations in transportation networks by combining and connecting different representations of traffic flows (microscopic, macroscopic) along with the corresponding spatial representations (links, nodes at different levels of aggregation) and temporal scales (discrete time steps) in a single framework. Multi-scale modelling is an increasingly important research subject, particularly in the design and evaluation of large scale ITS, since these may affect travel and driving behaviour at different scales, ranging from route choice behaviour to (ontrip) path planning and actual driving behaviour. In such applications, the consistency of travel and traffic operations between the different models in a multi-scale framework is critically important. In this paper we provide a new taxonomy that enables the identification of the sources of inconsistency between traffic flow models of different scales. We show that these inconsistencies can be categorised into two main categories, which we term global and local inconsistency respectively. Since these two classes relate to the endogeneity or exogeneity of the variables and parameters of interest, two models connected in a multi-scale framework can be both locally and/or globally inconsistent. On the basis thereof we sketch two main solution directions targeted at resolving both classes of inconsistencies. In this paper we demonstrate the applicability of both solution directions with an example case. We close with a discussion on the further development of a multi-scale modelling framework based on these results.

It has been proposed that a macroscopic fundamental diagram (MFD) can be used as input for perimeter control strategies. We consider a network consisting of a perimeter with traffic lights and a subnetwork without traffic lights. The MFDs for the two parts, i.e. the controlled subnetwork and its perimeter, are determined by means of microsimulation. We found that different signal settings change the shape of the MFD for both parts considerably. Also the ratio between the performance of the subnetwork and the corresponding perimeter is often fairly constant regardless of the signal timings. Thus, metering traffic heading towards the subnetwork is not always a good control strategy, as the performance of the subnetwork is eventually affected in the same way as the performance of the perimeter. Furthermore it has been found that the shape of the MFD of the perimeter, including the critical network density, strongly depends on the signal timings. It is therefore concluded that the MFD is difficult to use for control strategies aiming to adapt signal timings, because the changed signal timings themselves will result in a different value for the critical network density.

This paper presents a framework for real-time integrated and anticipatory traffic management. It uses model predictive control including route choice to optimize traffic management. It is tested for one small network with good results. Simulations varying the prediction horizon, the number of iterations of the assignment within the optimization loop and the optimization criterion were conducted and results of these simulations are given. The framework can be used as a basis for further research on this topic.

The Praktijkproef Amsterdam (Field Operational Test Integrated Network Management Amsterdam) is one of the first large-scale FoTs testing coordinated network wide deployment of traffic management in practice. After a successful Proof-of-Concept (PoC) in 2009, in 2013 the first parts of the concept approach will be tested in the field, working towards a full deployment from 2014 onwards. This paper describes the hierarchal control approach that has been developed in order to operationalise the control paradigm developed during the PoC. The paper explains this control framework, the functional architecture, and the main control algorithms. The paper also shows the first results of the application of the control approach in a simple test network.

Blocking back is one of the major causes of degradation of freeway network performance. When at a freeway diverge one of the outgoing links is jammed, the traffic heading to the other outgoing link will often also be delayed due to the back-propagation of the jam tail into the common area upstream of the diverge. Most approaches consider this as a secondary effect of the real bottleneck on the congested link, and use this as a motivation to try to resolve the bottleneck. While this is a valid approach, in some cases this is not possible, and the traffic heading to the non-congested outgoing link will suffer from a bottleneck that is not even on their route. Here we solve the blocking back problem by separating the traffic toward the two outgoing links by dynamic lane allocation. We analyze the jam propagation and traffic state evolution for the cases with and without dynamic lane separation. Next we develop two control approaches, one based on the queue tail location and the other based on the shock wave theoretic considerations. The two controllers are evaluated by microscopic simulations for their resulting performance in terms of total travel time. The simulation results indicate that the queue tail location based controller is superior to the shock wave theory based controller.

Traffic is composed of different vehicle classes, each characterized by the vehicle length, the value of time (VOT) and other class-specific properties. Multi-class traffic flow models aim to capture these class-specific differences. In this paper, a multi-class model is appended with a monetary cost framework accounting for the value of time of each vehicle class, thereby enabling the computation of the total network costs. Subsequently, the framework is analytically applied to show the conditions under which it is justified to prioritize certain traffic classes in order to maximize the monetary flow (which is equivalent to minimizing the total monetary network costs). An illustrative example is given in which the framework is used to find the conditions (bottleneck capacity, traffic demand, pce, VOT) under which activating a dedicated lane for trucks is beneficial.

Recently, an optimal control framework has been put forward to model human driver behavior and Advanced Driver Assistance Systems (ADAS). Although the models are shown to be face valid, applications are limited to the open-loop optimal control problem and assume that a driver expects a certain dynamic speed profile of his predecessor. While human drivers are assumed to be good in anticipating traffic, ADAS are poor at predicting the dynamics of other vehicles, due to limited looking-ahead range and errors of sensors and actuators.

In this paper we explore the idea of dimensionality reduction and approximation of OD demand based on principal component analysis (PCA). First, we show how we can apply PCA to linearly transform the high dimensional OD matrices into the lower dimensional space without significant loss of accuracy. Next, we define a new transformed set of variables (demand principal components) that is used to represent the fixed structure of OD matrices in lower dimensional space. We update online these new variables from traffic counts in a novel reduced state space model for real time estimation of OD demand. Through an example we demonstrate the quality improvement of OD estimates using this new formulation and a so-called ‘colored’ Kalman filter over the standard Kalman filter approach for OD estimation, when correlated measurement noise is accounted due to reduction of variables in state vector.

Uphill sections have often been identified as capacity bottlenecks in freeway networks. One of the main reasons seems to be that drivers reduce speed when they reach the beginning of an uphill section. With high traffic demand, the deceleration of the first vehicle of a platoon can generate a flow disturbance that amplifies as it propagates upstream, triggering the formation of a traffic jam. This paper presents a proof of concept by exploring whether equipping the leader of a platoon with an in-vehicle Gradient Compensation System (GCS) can improve traffic flow efficiency on uphill sections. The GCS assists the driver in performing the longitudinal driving task at uphill sections. We present the results of a series of traffic simulation experiments in which a platoon of vehicles drive on a single-lane freeway stretch containing an uphill section. The phenomenon of speed reduction is modeled by means of a sub-microscopic traffic simulation program. The results show that if the platoon leader is not equipped with the GCS, its speed drop at the beginning of the uphill section can cause a traffic breakdown, as observed in reality. However, if the platoon leader is equipped with the GCS, the magnitude of the speed drop is reduced, preventing congestion formation.

Ramp Metering is often applied in Dynamic Traffic Management (DTM) to control the inflow onto freeways. Although traffic usually consists of different vehicle classes such as cars and trucks, these conventional ramp meter installations operate on all classes equally, thereby disregarding the traffic composition. This paper introduces a multi-class ramp meter (MCRM) which controls the traffic vehicle-class specifically. It regulates the inflow to ascertain a specified inflow composition. In this contribution, the control algorithm of multi-class ramp metering is presented. Furthermore, the effects on travel time, queue length and total delay are outlined.

Both from theory and from practice it is known that the variable speed limit algorithm SPECIALIST can resolve moving jams on freeways. However, the algorithm does not take on-ramps into account in its current form, while on-ramp traffic may have an impact of the effectiveness of the algorithm. In this paper we extend the theory of SPECIALIST to include on-ramps and ramp metering. The theory of SPECIALIST is formulated in the framework of shock wave theory and it results in a control scheme that describes the propagation of the fronts (boundaries) between the various traffic states, and describes the time and location where the speed limits need to be active. Here we extend this with the traffic states describing the additional flow from an on-ramp (either metered or not), and develop the resulting control schemes. Since the location of the on-ramp relative to the moving jam and to the control scheme can be categorized into four cases, we develop four different control schemes. The developed control schemes can be used to analyze offline traffic data to assess the number of moving jams that can be resolved. Resolvability basically depends on (1) whether the traffic situation (speeds, flows, etc.) is such that a control scheme can be constructed, and (2) whether the control scheme physically fits on the freeway stretch equipped with variable speed limit displays. As an example, we analyze the data of a stretch on the A2 freeway in the Netherlands for various speed limit values and maximum on-ramp flows. The results show that for the number of resolvable moving jams the fact that the on-ramp flows are taken into account is more important than the metered value of the on-ramp flow.

Traffic management on a network level is not only theoretically a complex problem, but it also involves the realization of the road authorities' traffic management policy. In the Netherlands, this policy harmonizes the interests of involved stakeholders by means of a common vision upon the network functioning expressed in road priorities and corresponding desired service levels. This paper presents a route guidance control methodology that is able to take the formulated policy into account. This is achieved by degrading and restoring service levels of routes according to their difference in importance with respect to network performance. The proposed method is the first dynamic routing approach that is able to operationalize the Dutch traffic management policy in line with how it is formulated. The proposed methodology consists of a finite-state machine that determines the desired service levels based on predicted traffic conditions. These service levels are used in a feedback controller as setpoints for the desired travel times, resulting in the corresponding output signal of Variable Message Signs. By means of a test case, the performance of the method is compared with a controller that realizes a user equilibrium and an uncontrolled scenario. The evaluation identifies the potential of route control based on service levels when dealing with differently prioritized routes.

In the evacuation problem, departure time, route, and destination instructions are optimized to increase the effectiveness of the evacuation (i.e., more arrivals at the destinations). In literature, evacuation instructions are mostly optimized for a problem without uncertainty (the so-called nominal problem): one specific scenario regarding the system, the hazard and the evacuees is assumed. In this paper, the robust evacuation problem is formulated to create instructions dealing with the uncertainty. So-called scenarios are used that are specific representative realizations of the various uncertainties. The effectiveness of a set of instructions is based on the worst-case performance. By solving the problem, a guarantee can be given on the number of arrivals, either absolute or relative to the number of arrivals in the solution for the nominal problem. An earlier presented optimization method is extended to solve the robust problem. In a case study, the robust evacuation problems are solved under system, hazard, demand, and behavior uncertainty. The results show that for the specific case study, the guarantee on the number of arrivals by applying the relative robustness evacuation instructions is equal to 96.1%, compared to a guarantee of 85.3% in case that only nominal problems are solved.

Adverse conditions have been shown to have a substantial impact on traffic flow operations with a substantial impact on network performance. In order to quantify adaptation effects in driving behavior underlying this impact car-following models can be used. In this contribution it was examined to what extent the position of action points in the relative speed — spacing plane as well as acceleration and ‘jumps’ in acceleration in psycho-spacing models is influenced by perception of an incident in the other driving lane. Using two datasets and a new data analysis technique in order to determine the position of the action points, it was concluded that perception of an incident in the other driving lane significantly influences this position. Furthermore it was concluded that acceleration as well as the ‘jumps’ in acceleration at the action points differed substantially between the two datasets. Drivers reacted with smaller accelerations and larger jumps in acceleration to relative speed at the incident site. The fact that the positions of action points in the relative speed-spacing plane as well as the relationship between relative speed, spacing and (jumps in) acceleration differ substantially between the datasets leads to the conclusion that deterministic perceptual thresholds in the original formulation of psycho-spacing models do not hold in reality. It is recommended to develop a data-driven stochastic model based on the principles of psycho-spacing theory, able to describe adaptation effects in case of adverse conditions.

Freeway traffic state estimation is one of the central components in real-time traffic management and information applications. Recent studies show that the classic kinematic wave model can be formulated and solved more efficiently and accurately in Lagrangian (vehicle number-time) coordinates. This paper investigates the opportunities of the Lagrangian form for state estimation. The main advantage for state estimation is that in Lagrangian coordinates, the numerical solution scheme is reduced to an upwind scheme. We propose a new model-based extended Kalman filter (EKF) state estimator where the discretized Lagrangian model is used as the model equation. This state estimator is applied to freeway traffic state estimation and validated using synthetic data. Different filter design specifications with respect to measurement aspects are considered. The achieved results are very promising for subsequent studies.

In this paper we propose a hybrid motorway travel time prediction framework aimed at providing pre-trip travel information in case of roadworks. The framework utilises a first order macroscopic traffic flow model to predict the consequences in travel time of changes in both traffic demand and roadway capacity. Data-driven approaches are used to estimate both demand and capacity. From a large database of detailed loop detection data the most likely demand profiles are estimated for the considered workzone location, taking into account the possible effects of mobility management. Capacity estimation techniques are employed which combine historical data with likely capacity reduction factors as a consequence of the roadworks. On the basis of evaluation on a workzone case on a densely used 20 km stretch of a three lane motorway in The Netherlands, it is demonstrated that the proposed approach is able to predict travel time within a 20% margin in over 96% of all cases. The proposed approach is suitable for trip planning purposes (e.g. route planners and travel information websites), and may also be utilised by the road authorities in the actual planning of the roadworks.

Adverse weather conditions have been shown to have a substantial impact on traffic flow operations. It is however unclear which adaptation effects in actual longitudinal driving behavior underlie this impact, how these adaptation effects relate to freeway capacity as well as to what extent current mathematical models of car-following behavior are adequate in incorporating these adaptation effects. In this regard a driving simulator experiment with a repeated measures design was performed in order to examine the influence of fog on adaptation effects, freeway capacity and parameter value changes and model performance of the Helly model and Intelligent Driver Model. From the results followed that fog led to a decrease in speed as well as in acceleration. Furthermore a substantial increase in distance to the lead vehicle was observed. These effects were implemented and simulated in a traffic simulation model. A substantial reduction in freeway capacity was found. This stresses the need to possess models of driving behavior, which are adequate in describing and predicting these adaptation effects. From the estimation results of the Helly model and IDM using a calibration approach for joint estimation followed that sensitivity factors, maximum acceleration and deceleration decreased substantially after the start of the adverse weather condition. Parameters representing headway increased significantly. Furthermore it followed from the results that the estimated models decreased in performance after the start of the adverse weather conditions

Freeway traffic state estimation is crucial for dynamic traffic management (DTM), Advanced Traveler Information Systems (ATIS) and highway performance analyses. Raw data collected by dual-loop detectors or GPS devices provide information about flow and speed at points in space and time. However, these observations are noisy and incomplete. The Adaptive Smoothing Method (ASM) estimates the traffic state between the observation points and reduces the noise inherent to observations. Current implementations of the ASM apply its model in a straight-forward manner, which leads to high computation times. In this paper, two new implementations are developed that drastically reduce the computation time while preserving the estimation quality. In the first implementation, the ASM is discretized to apply the cross-correlation. This is based on matrix operations, which are efficiently implemented and fast in execution. In the second implementation, the ASM is reformulated to apply the Fast Fourier Transform (FFT). The FFT, too, is based on fast matrix operations. These two new implementations are sequential programs, containing no loops. Experiments with a setup used in practical applications and real data show computation times of just a few seconds. These are computation time improvements of two orders of magnitude. The rapid computation of the traffic state makes the ASM with the proposed implementations applicable for real-time applications.

We present the real-world test of the SPECIALIST algorithm in which dynamic speed limits were used to resolve shock waves on freeways. The real-world test was performed in the period September 2009–February 2010 on a 14 km long stretch on the Dutch A12 freeway.

This paper proposes a new method called TravRes for accurate reconstruction of high-resolution (e.g. 500m*60sec) time-space speeds on the basis of sampled low-resolution floating car data. The new method is superior to naive methods, methods which reconstruct time-space speeds by simply dividing travel distance by travel time between consecutive FCD samples. The new method uses an iterative trajectory reconstruction technique and data consistency approach, and as a result improves estimation accuracy by up to 90% relative to the naive method, particularly at low speeds. In terms of estimated travel times (from these reconstructed speeds), TravRes can improve the accuracy by up to 5 times.

Accurate and reliable speeds measured with local traffic sensors are critically important for many applications, such as travel time estimation, state estimation and prediction and dynamic traffic control. Many local sensors (e.g. the dual induction loops used in The Netherlands, UK and in Germany) calculate average speed by arithmetic time averaging, which leads to strongly biased estimates. This paper proposes a speed-bias correction algorithm based on notions from first order traffic flow theory and empirical flow-density relationships, using only a limited number of parameters. On the basis of a microscopic simulation study, it is demonstrated that this algorithm is able to correct the speed bias due to time averaging considerably. A more extensive version of this paper can be found at http://dux.ict.tbm.tudelft.nl/Research/SpeedCorr.pdf

The steadily increasing numbers and lengths of traffic jams on freeways have led to the use of Dynamic Traffic Management (DTM) measures all over the world. Ramp metering control has proven to be one of the most efficient means to reduce freeway congestion. It is expected that integrated and coordinated application of DTM measures will further improve its impact. This paper studies a new coordinated ramp metering control algorithm called HERO/RWS. This algorithm has been developed for the current Dutch ramp metering systems and it will be applied on the Amsterdam A10 freeway network in the near future. The aim of this algorithm is to postpone congestion on freeways by effectively using ramp storage space from upstream on-ramps. The control scheme is simple and real-time operable. VISSIM-based microscopic simulation results show that the HERO/RWS coordinated control outperforms non-coordinated control. This control algorithm turns out to provide less congestion, higher mean speeds and lower travel time spent on the freeway.

The analytical evacuation network model EVAQ enables forecasting traffic flow conditions on a road network for a wide range of emergency situations, such as hurricanes, bushfires and floods. The proposed model is innovative as it combines voluntary evacuation, with possible pre-trip and enroute decisions, and mandatory evacuation, with prescribed destinations, routes and departure times, within a single generic model framework. Furthermore, we introduce dynamic road infrastructure, in which characteristics such as speed limits, capacity and flow direction can be time-varying due to the hazard's progress in space and time and prevailing traffic regulations and control measures. The evacuation model can function as a decision support system for regional authorities and emergency services supervising an evacuation or constructing an evacuation strategy for planning purposes.

Modeling realistic driving behavior at signalized intersections is crucial for many applications, for instance to determine the traffic signal performance, to assess the effect of different control strategies, or to estimate traffic emissions. In these applications, often microscopic models are used to simulate the trajectory of each vehicle. Despite the possibility to model vehicles with great detail and at fractions of a second, speed, acceleration and deceleration characteristics are determined by parameters that are rarely calibrated using real data, and default parameters are often chosen. This is because collecting real vehicle trajectories near traffic signals is a challenging task. This paper presents a method to collect such dataset using image processing techniques. This methodology allows one to obtain vehicle trajectories near a signal control, and to measure individual vehicles speeds and accelerations at a microscopic level. We focus on the analysis the empirical distributions of speeds and accelerations observed with this unique dataset near and up to a few meters upstream of the stop-sign. We compared these distributions with the results of repeated simulations of two microscopic software programs, using default parameters. Some inconsistencies were found with this comparison, which suggests that the two analyzed microscopic simulation programs run with default parameters do not provide realistic results for this type of road sections.

In the Netherlands, a field operational test was conducted in 2006 to assess the impact of two advanced driver assistance systems (ADAS), namely adaptive cruise control (ACC) and lane departure warning (LDW) systems. The research goal was to estimate the effects of these systems on road capacity, safety and emissions. In this paper we focus on the interaction between driver and ACC system using the data from this FOT. It is found that drivers choose headway settings according to their manual driving behavior. Moreover, they often keep the system deactivated under dense traffic conditions. It is also observed from the data that the system, once de-activated, either automatically or manually, needs some time to become again active. These findings imply that, even with 100% cars on the road equipped with ACC, manual driving behavior will still be a determinant factor.

Microscopic simulation models have become widely applied tools in traffic engineering. Nevertheless, parameter identification remains a difficult task. This is for one caused by the fact that parameters are generally not directly observable from common traffic data. The second difficulty stems from the fact that real driving behavior is variable in time and space, etc. This paper puts forward a new approach to identify changing parameters of delayed car-following models, i.e. models that include a reaction time. The approach is based on the unscented particle filter approach, which is generalized to enable estimation of reaction times. The estimation of this true delay is achieved without linearization. Besides the methodological contribution, we show empirical evidence for changing driving behavior by applying the approach to real-life microscopic traffic data

This paper discusses the importance of spillback modeling in assessing the impacts of dynamic traffic management. Specifically, the paper compares the influence of route information in a simulation with spillback and without spillback modeling. The considered case entails an evacuation of a city (circa 250,000 cars). The results indicate that in the non-spillback simulation, the network performance is indifferent for route measures, whereas in the more realistic simulation with spillback, there are differences between different types of information provision. It is concluded that a traffic model without simulation of spillback not only gives an underestimation of evacuation times - which is obvious -, but more importantly it completely underestimates the effect that can be obtained by route information. The travel time gain according to the non-spillback model is 18%, whereas the spillback model predicts 17% reduction of travel time. It is also shown that for the case at hand, the congestion caused by spillback can be avoided by providing appropriate information

Compared to the motorway network, rural and urban roads are very unsafe. Advanced Driver Assistance Systems can be used to increase traffic safety, thereby improving the conditions for effective deployment of the underlying network for integrated traffic management. In this study, we have identified ADA systems that are expected to have a high impact on traffic safety by coupling characteristics of accidents to the functionality of different ADA systems. Based on the results of this study, Intelligent Speed Adaptation (ISA) and Intersection Crash Avoidance systems appeared most likely to substantially improve the safety on non-motorway roads. ISA was selected for further analysis due to the likelihood of timely deployment of the system. Microscopic simulation model was used to quantify the impacts on traffic safety for an urban highway with controlled intersections near the Dutch city of Delft. Depending on the penetration level, the results of the simulation study showed that the decrease in the total number of accidents ranges from 4% to 19%.