Large-scale Urban Networks Research Articles

Data fusion for estimating Macroscopic Fundamental Diagram in large-scale urban networks

Since the concept of the Macroscopic Fundamental Diagram (MFD) has been introduced, many studies have investigated the existence and characteristics of the MFD using empirical and simulation data. MFD is a powerful and efficient model for monitoring and managing large-scale urban networks. However, estimating the MFD for large-scale networks faces important challenges; monitoring resources are often limited in such networks. Furthermore, common sensors that are used to collect traffic data (i.e., loop detectors and probe vehicles), have limitations of their own. For instance, loop detectors are fixed sensors and cannot provide accurate density measurements. On the other hand, to estimate the MFD using probe vehicle data, the probe penetration rate must be known a priori. Given that the individual sensors cannot provide complete and accurate traffic measurements, combining the traffic data from multiple sources may improve the estimation of the MFD. The aim of this study is to combine two traffic data sources to estimate the MFD for a large-scale urban network, where the distribution of probe vehicles across the network is not necessarily homogeneous. The two traffic data sets used in this study are probe vehicle data with an unknown penetration rate, and full-scale approximate traffic data which is produced based on loop detector data. We compare the results of the fusion method with the results of a baseline method, which only uses loop detector measurements. The average flow and density estimations resulting from the Bayesian fusion method outperform the baseline method. We observe a particularly significant improvement in average density estimations, which reaffirms that loop detectors cannot accurately measure the average density.

Dynamic density and flow reconstruction in large-scale urban networks using heterogeneous data sources

This work proposes an estimator for the vehicle density of every road section of a large urban traffic network using a data-based approach. We assume that a limited number of flow and turning ratio sensors can be installed and that aggregate Floating Car Data are available. To validate the estimator, real data obtained in the city of Grenoble in France is used. The results show that the trajectories of the real and estimated flows and density are very close, showing the good performance of the proposed methods.

Glocal (Global/Local) Control for Urban Road Traffic Networks

This paper provides a way of hierarchically distributed control for large-scale urban road traffic networks based on the concept of Glocal (Global/Local) Control. The proposed system consists of demand distribution control by dynamic route guidance (global) and traffic signal control (local), and they are collaborated each other through so called glocal adaptor to achieve increase of traffic volume (global objective) and jam alleviation (local objective). After explaining the structure of the glocal adaptor, we provide model predictive control-based algorithms for demand distribution control and distributed traffic signal control, which are completely independent. The effectiveness of the method is verified by numerical experiments.

Identifying traffic clusters in urban networks based on graph theory using license plate recognition data

Monitoring traffic states of urban road networks is essential to relieving traffic burdens and designing traffic management strategies. However, most traffic state estimation methods are data-driven and barely consider the traffic network topology characteristics. In this paper, we propose a model-driven approach to partition the network into several traffic clusters, providing a macro-perspective for traffic state monitoring and potential support for perimeter controls. Traffic state and network topology information are both considered in the approach. We obtain the traffic state index of each segment with license plate recognition data and traffic network topology information, considering the heterogeneity and direction of segments. The road graph is constructed based on graph theory, and we modify the RatioCut algorithm and turn hyper-parameters automatically to identify traffic clusters in the road graph. We verified the proposed approach by a large-scale urban network in Hangzhou, China. The visualization and comparison results with conventional spectral clustering, DBSCAN, and K-means demonstrate the superiority and application value of the proposed approach.

Real-time turning rate estimation in urban networks using floating car data

Real-time information about the dynamic variation of turning rates at intersections is essential for achieving a more efficient traffic control and management system. Limitations in obtaining measurements of turning movements have drawn attention to the need for alternative methods for estimation of turning rates. In a large-scale urban network, characteristics such as disparate signal plans and various intersection layouts make defining a generic model for the entire network challenging. For this reason, most of the proposed methods are impractical for real-field implementation. The application of existing optimization-based methods is also limited by computational complexity, the need for pre-defined assumptions and insufficient stationary traffic measurements (e.g. loop detectors and plate recognition cameras) in urban networks. In this study, we take advantage of probe vehicles to propose a hybrid method that noticeably reduces the need for stationary measurements in the network. This hybrid method consists of two core stages: signal processing, and physical data-driven modeling to consecutively estimate turning rates. In the signal processing phase, we use wavelet transformation to extract fundamental components of inferred floating car turning rates. These components are then used as inputs to a turning rate estimation model. Using these decomposed components of floating car turning rates instead of raw input helps to balance the effect of low and high-frequency signal components and leads to a more accurate estimation. Investigation of the proposed method in a microscopic simulation environment reveals the ability of the hybrid model to detect turning rate variations even when a severe incident occurs.

An extended Kalman filter approach for real-time state estimation in multi-region MFD urban networks

The problem of traffic state estimation for large-scale urban networks modeled with MFD dynamics is studied here. Given a network partitioned in a number of regions, aggregated traffic dynamics describe the vehicle accumulation in each region, as well as transfer flows to and from neighboring regions. Considering that MFD accumulation-based models have been integrated in perimeter control approaches, this work tackles the real-time estimation problem when limited data is available. An estimation engine is developed according to the Extended Kalman Filter (EKF) theory; it seeks to estimate the real state of the multi-region dynamic system based on traffic sensors’ measurements. First, a stochastic model is presented for the dynamics of the process (plant). Then, the EKF estimation scheme is described based on a simpler aggregated model of dynamics and some real-time measurements. Estimation accuracy is investigated through detailed micro-simulation of downtown Barcelona by studying a realistic configuration of real-time measurement availability through loop detector data; however, the developed methodology is generic. The state vector we seek to estimate, as well as the available measurements configuration, can be altered according to the application. The proposed methodology is tested both in macro- and micro-simulation; resulting estimated traffic states (i.e., regional accumulations, demands, and distribution of outflows) are compared to actual ones obtained from the stochastic plant. The developed algorithm can be utilized by closed-loop online urban traffic management strategies to feed the estimated traffic state back to the controller.

Decentralized store-and-forward based strategies for the signal control problem in large-scale congested urban road networks

Signal control strategies for congested urban road networks designed in a centralized framework require many communication links, serious processing power, and infrastructure for the centralized coordination. As a result, strategies based on a centralized framework are not scalable. The use of decentralized signal control strategies for large-scale urban traffic networks is a solution to this problem, since it allows for the implementation of such strategies on networks whose centralized solution is not easily scalable. This paper addresses the problem of designing a decentralized traffic-responsive signal control solution, proposing two methods based on different formulations of the store-and-forward model: (i) the Decentralized Traffic-responsive Urban Control (DTUC) method; and (ii) the Decentralized Decoupled Traffic-responsive Urban Control (D2TUC). The decentralized configuration is such that each intersection is associated with one computational unit, with limited computational power and memory, which controls the traffic signals of the incoming links. Sufficient conditions for the controllability of the considered store-and-forward models are also presented. Both methods are validated resorting to numerical simulations of the urban traffic network of Chania, Greece, for two demand scenarios, and their performance is compared with the performance of the Traffic-responsive Urban Control (TUC) centralized strategy. One of the proposed decentralized methods, D2TUC, is shown to match the performance of TUC.

Modeling, estimation, and control in large-scale urban road networks with remaining travel distance dynamics

City-scale control of urban road traffic poses a challenging problem. Dynamical models based on the macroscopic fundamental diagram (MFD) enable development of model predictive perimeter control methods for large-scale urban networks, representing an advanced control solution carrying substantial potential for field implementation. In this paper we develop a multi-region approximation of the trip-based model, which describes in more details the trip length characteristics compared to existing accumulation-based models. The proposed M-model includes effects of the total remaining travel distance on the transfer flows driving the vehicle accumulation dynamics, potentially yielding improved accuracy over the standard production-over-trip length approximation of the outflow MFD considered in many works on MFD-based modeling and control. We explain that to properly perform perimeter control, boundary queue dynamics have to be integrated. Furthermore, model-based parameter estimation (MBPE), nonlinear moving horizon observer (MHO), and model predictive control (MPC) formulations for the proposed models are presented, forming an integrated traffic control framework. Microsimulation-based case studies, considering an urban network with 1500 links, where the model parameters obtained by MBPE method are used in MHO and MPC design, demonstrate the efficient operation of the proposed framework.

Data Augmented Deep Behavioral Cloning for Urban Traffic Control Operations Under a Parallel Learning Framework

It is indispensable for professional traffic signal engineers to perform manual operations of traffic signal control (TSC) to mitigate traffic congestion, especially with complicated scenarios. However, such a task is time-consuming, and the level of congestion mitigation heavily relies on individual expertise in engineering practice. Therefore, it is cost-effective to learn traffic engineers’ knowledge to enhance the problem-solving skills for a large-scale urban traffic network. In this paper, a data augmented deep behavioral cloning (DADBC) method is proposed to imitate the problem-solving skills of traffic engineers. The method is under a conceptual framework, parallel learning (PL) framework, that incorporates machine learning techniques for solving decision-making problems in complex systems. The DADBC method enhances a hybrid demonstration by exploiting a generative adversarial network (GAN) and then uses the deep behavioral cloning (DBC) model to learn traffic engineers’ control schemes. According to the validation results using the real manipulation data from Hangzhou, China, our method can imitate complex human behaviors in intervening traffic signal control operations to improve traffic efficiency in urban areas.

Resilient multivariable perimeter control of urban road networks under cyberattacks

The Macroscopic Fundamental Diagram (MFD) concept has become a very valuable tool for modeling the aggregated dynamic of large-scale urban traffic networks. This advance in traffic modeling has fueled the development of many urban traffic control (UTC) policies based on the MFD concept. In particular, perimeter control policies with the aim of maintaining a large-scale urban traffic network operating at its maximum capacity by manipulating the transfer flow of vehicles between adjacent homogeneous regions. In addition to achieve a satisfactory performance level, it is desirable that the control system would maintain a reliable operation under unanticipated events such as cyberattacks. It is important to note that modern UTC systems extensively rely on information technology infrastructures making them vulnerable to cyberattacks. Due to the occurrence of real-world cyberattacks affecting controlled physical systems in different fields, the threat of cyberattacks is becoming a serious and increasing concern for the proper operation of UTC. This paper presents a resilient multivariable control framework for large-scale urban traffic networks subject to several types of cyberattacks such as deception attacks and denial of service (DoS) attacks. Additionally, a feedforward–feedback control scheme combining a time-optimal controller with the resilient multivariable controller is also proposed. The aim of this feedforward–feedback controller is to steer the accumulations to their desired values after a large deviation. Finally, the performance of the proposed resilient controller is demonstrated by means of some realistic simulations of urban traffic networks, including stochastic errors in the traffic demands and in the MFD model, in addition to cyberattacks.

Temporal and Spatial Evolution of Passenger Flow in an Urban Rail Transit Network During Station Closure

In this study, we focused on the evolution of passenger flow in an urban rail transit network in both temporal and spatial dimensions under the event of station closure. We constructed an extended space-time-state hyper network so that we could utilize a better-defined three-dimensional solution space to describe passenger behaviors and pedestrian characters. The space-time-state network was extended by adding dummy arcs and setting the “block space-time domain” based on the original spatiotemporal network. Based on the extended space-time-state hyper network, a passenger flow evolution model that aimed to minimize the usage cost during station closure was established considering the strict capacity constraints (including corridor passing capacity, platform load capacity and train transport capacity). For a large-scale urban rail network, we developed a decomposition solution framework in which the Lagrangian relaxation and semi-assignment algorithms were adopted. A real-world instance was implemented based on the Beijing subway network with complete smart card data for each passenger for his/her origin and destination, while the specific space-time trajectories of all passengers and time-dependent passenger volumes in trains and transfer corridors were estimated. The results of a numerical experiment suggest that the proposed passenger flow evolution method can provide a rich set of passenger volume inferences for advanced transit planning and management applications, such as passenger flow control, passenger flow guidance, and real- time train scheduling.

Vulnerability Assessment of the Urban Rail Transit Network Based on Travel Behavior Analysis

Vulnerability assessment plays a crucial role in the management of complex and large-scale urban rail transit networks (URTNs). Even a temporary disruption may paralyze an entire public transportation network under a cascading failure. In the terms of vulnerability of the URTN, most studies have focused on changes in URTN topological characteristics and travel costs under disturbances, thereby neglecting the impact of alternative public transportation modes on passenger flow redistribution within the URTN. Therefore, in this study, considering the impact of changes in travel costs and modes on passenger flow redistribution, a novel model is proposed to analyze the cascading failure of weighted URTNs. Furthermore, adopting the estimated operation interruption time of the station (EOITS) to characterize the system disturbance intensity, the dynamic vulnerability of the URTN is simulated and analyzed. The results show that the EOITS exerts a major impact on the vulnerability of URTNs. With increasing of EOITS, the travel behavior of passengers becomes more diverse, and the process of cascading failure in URTNs also becomes more challenging to control. This research has implications for managers in terms of implementing of protective measures to mitigate the cascading failure risk and avert extreme catastrophes.

Simulation-based optimization of toll pricing in large-scale urban networks using the network fundamental diagram: A cross-comparison of methods

Simulation-based optimization (SO or SBO) has become increasingly important to address challenging transportation network design problems. In this paper, we propose to solve two toll pricing problems with different levels of complexity using the concept of the macroscopic or network fundamental diagram (MFD or NFD), where a large-scale simulation-based dynamic traffic assignment model of Melbourne, Australia is used. Four computationally efficient SBO methods are applied and compared, including the proportional-integral (PI) controller, regressing kriging (RK), DIviding RECTangles (DIRECT), and simultaneous perturbation stochastic approximation (SPSA). The comparison reveals that these methods work equally well on the simple problem without exhibiting significant performance differences. But, for the complex problem, RK manifests itself to be the best-performing method thanks to its capability of filtering out the numerical noise arising from computer simulations (i.e. allowing for non-smoothness of the objective function). While the PI controller is a more competitive solution to the simple problem given its faster rate of convergence, the poor scalability of the method in the complex problem results in limited applicability. Two caveats, however, deserve emphasis: (i) the chosen critical network density of the NFD does not necessarily represent a robust network control or optimization threshold, as it might shift in the presence of toll pricing; and (ii) re-interpolation is required as part of RK in order to achieve global convergence.

EvoTSC: An evolutionary computation-based traffic signal controller for large-scale urban transportation networks

Traffic Signal Control (TSC) is a crucial component in the modern intelligent transportation systems. Typically, the TSC can be formulated as a bilevel optimization problem, which is comprised of the signal timing and the traffic assignment modules. The problem is challenging that the existing approaches usually endure a huge computational cost. As a result, many TSC approaches focus on relatively simple and small transportation networks, which do not satisfy the practical situations. To address above issues, this paper proposes an evolvable TSC (EvoTSC) system, which adopts nature-inspired techniques to realize the global optimization of the TSC in large-scale urban transportation networks. Particularly, it involves two evolutionary computation components. The first component is an Adaptive Differential Evolution (ADE) to optimize the signal timing. Meanwhile, the traffic assignment process is included in the solution evaluation of the ADE to react to the traffic flow dynamics. The second component is an off-line Niching Ant Colony Optimization (NACO), which aims to provide the traffic assignment with sets of multiple promising routes beforehand. This way, the EvoTSC system avoids repeatedly building candidate routes for the traffic assignment, which can greatly save the computational cost of the ADE to evaluate each solution in a large-scale transportation network. In experiments, we carry out comparisons of different TSC approaches on both synthetic and practical transportation networks. The experimental results validate the effectiveness of the proposed EvoTSC system.

A methodology for identifying critical links and estimating macroscopic fundamental diagram in large-scale urban networks

The Macroscopic Fundamental Diagram (MFD), which exhibits the relationship between average flow and average density of an urban network, is a promising framework for monitoring and controlling urban traffic networks. Given that monitoring resources (e.g. loop detectors, probe vehicle data, etc.) are limited in real-world networks, acquiring adequate data to estimate an MFD is of crucial importance. This study presents a novel, network-wide approach to identifying critical links and estimating average traffic flow and density. The proposed model estimates the MFD using flow and density measurements from those critical links, which constitute only a small subset of all the links in the network. To find the critical links, we rely on historical probe vehicle data, and propose a model that builds on Principal Component Analysis (PCA), a dimensionality reduction and a feature selection method. Essentially, using PCA, a large number of possibly interrelated variables in a dataset can be reduced to a set of smaller uncorrelated variables, while maintaining as much information as possible in the dataset. The resulting uncorrelated variables, or the principal components, indicate the major patterns or the dominating features of the original dataset. Additionally, PCA enables the (approximate) reconstruction of the full-scale dataset from the selected features (or principal components). In this work, we apply PCA in order to identify the main traffic features from a probe vehicle dataset; then, we find the links that are associated with these features (i.e., critical links); then, we locate loop detectors on those links to collect flow and density data; and finally, we reconstruct the full-scale data, building on the PCA mechanism. This gives us the flow and density of all links, from which we can effectively estimate the MFD.

Recursive Hybrid Heuristic Algorithm for Routing a Track-Geometry Car through a Large-Scale Urban Subway Network

AbstractRail track inspection is essential to urban rail infrastructure management in terms of safety. The track geometry car is one of the important inspection vehicles for track condition measure...

Effects of controlling aggressive driving behavior on network-wide traffic flow and emissions

Abstract This paper assesses the impact of adopting smooth driving habits on traffic and emissions in large scale urban networks. An integrated data driven and simulation methodology is proposed, which, first tries to reveal the prevailing driving profiles using real world driving behavior data gathered from smartphones, and second, to simulate the effect of controlling the observed profiles on traffic and emissions using microscopic simulation. The available database contains a total of 4156 urban trips originating from 100 distinct drivers, for who detailed driving behavior data exist. A k-means clustering algorithm is implemented to extract the driving profiles based on speed and acceleration data. The simulation results show that smooth driving leads to a statistically significant reduction in the emissions of the most important air pollutants. Moreover, limiting the variability of the acceleration to a narrower range, as it occurs in the eco profile, leads to an increase in the output of vehicles in comparison to the other profiles.

Toward A Thousand Lights: Decentralized Deep Reinforcement Learning for Large-Scale Traffic Signal Control

Traffic congestion plagues cities around the world. Recent years have witnessed an unprecedented trend in applying reinforcement learning for traffic signal control. However, the primary challenge is to control and coordinate traffic lights in large-scale urban networks. No one has ever tested RL models on a network of more than a thousand traffic lights. In this paper, we tackle the problem of multi-intersection traffic signal control, especially for large-scale networks, based on RL techniques and transportation theories. This problem is quite difficult because there are challenges such as scalability, signal coordination, data feasibility, etc. To address these challenges, we (1) design our RL agents utilizing ‘pressure’ concept to achieve signal coordination in region-level; (2) show that implicit coordination could be achieved by individual control agents with well-crafted reward design thus reducing the dimensionality; and (3) conduct extensive experiments on multiple scenarios, including a real-world scenario with 2510 traffic lights in Manhattan, New York City 1 2.

Unraveling reaction-diffusion-like dynamics in urban congestion propagation: Insights from a large-scale road network

We study the dynamical process of congestion formation for large-scale urban networks by exploring a unique dataset of taxi movements in a megacity. We develop a dynamic model based on a reaction and a diffusion term that properly reproduces the cascade phenomena of traffic. The interaction of these two terms brings the values of the speeds on road network in self-organized patterns and it reveals an elegant physical law that reproduces the dynamics of congestion with very few parameters. The results presented show a promising match with an available real data set of link speeds estimated from more than 40 millions of GPS coordinates per day of about 20,000 taxis in Shenzhen, China.

Integrated Control for a Large-Scale Mixed Network of Arterials and Freeways

An urban traffic network is typically composed of two subnetworks: freeways and arterials. When the exchange traffic flow between these two subnetworks is over concentrated on an off-ramp or on-ramp, local congestion occurs in the traffic network. We propose an integrated control (IC) method to effectively adjust and coordinate the traffic flow between the two subnetworks of a large-scale urban traffic network. The proposed method depends on a subregion partitioning scheme based on the macroscopic fundamental diagram (MFD) and the road capacity constraints and integrates macroscopic traffic guidance, ramp-coordinated (RC) control and MFD subregion perimeter control to adjust the traffic flows for exchange between the two subnetworks. A solution method for the proposed IC model is designed using a Lagrange-quasi-Newton algorithm. Simulation studies of a congestion scenario are presented, and the results indicate that in comparison with no RC control, RC control only and a combined RC and perimeter control scheme, the proposed control scheme can reduce the delay, trip distance and trip time for the entire network.