At an urban intersection, traffic performance is highly influenced not only by signal timing schemes but also by lane assignments. However, dynamic adjustment of lane configurations remains understudied due to the prevalent use of predetermined lane arrangements. To address this research gap, we propose the TransLane framework, which collaboratively optimizes time and space resources within an intersection by implementing dynamic lanes. We model each intersection as an agent and apply a hierarchical recommendation approach to these agents. Additionally, the framework addresses the challenge of training an agent in the absence of sufficient data by using transfer learning techniques. Specifically, a new intersection agent can be trained from pre-trained agents via model and sample transfer mechanisms, which facilitate the reuse of traffic control knowledge from similar intersections. The efficacy and superiority of TransLane are demonstrated through simulation studies that show it outperforms models that optimize space or time resources independently using widely adopted traffic control optimization methods.

Urban traffic control is a multifaceted and demanding task that necessitates extensive decision-making to ensure the safety and efficiency of urban transportation systems. Traditional approaches require traffic signal professionals to manually intervene on traffic control devices at the intersection level, utilizing their knowledge and expertise. However, this process is cumbersome, labor-intensive, and cannot be applied on a large network scale. Recent studies have begun to explore the applicability of recommendation system for urban traffic control, which offer increased control efficiency and scalability. Such a decision recommendation system is complex, with various interdependent components, but a systematic literature review has not yet been conducted. In this work, we present an up-to-date survey that elucidates all the detailed components of a recommendation system for urban traffic control, demonstrates the utility and efficacy of such a system in the real world using data and knowledge-driven approaches, and discusses the current challenges and potential future directions of this field.

In urban environments, traffic flow can display a wide range of flow patterns and turning ratios, necessitating the strategic allocation of spatio-temporal resources to improve intersection efficiency. While a significant body of research has been dedicated to signal timing optimizations via model-based or learning-based methodologies, there is a noticeable lack of studies that investigate the modification of lane configurations to maximize space resources in this setting. Moreover, model-based algorithms can be computationally demanding and difficult to train in the absence of ample data. To tackle these issues, we introduce Trans-Sig, a transfer learning recommendation framework designed to aid decision-support in intersection traffic control. Trans-Sig capitalizes on dynamic lanes, a recent innovation aimed at promoting more efficient utilization of space resources. Adopting a Hierarchical Reinforcement Learning (HRL) strategy, Trans-Sig employs two reinforcement learning agents to address this co-optimization challenge. To enable the application of traffic control knowledge from homogeneous intersections under similar traffic flows, Trans-Sig integrates pre-trained agent pools and offers a pipeline for knowledge transfer from pre-trained to target intersections via model and sample supervision mechanisms. The effectiveness and superiority of Trans-Sig are validated through simulated intersections under various flows, where it outperforms models that optimize space or time resources independently.

Forecasting air pollution is an important activity for developing sustainable and smart cities. Generated by various sources, air pollutants distribute in the atmospheric environment due to the complex dispersion processes. The emerging sensor and data technologies have promoted the development of data-driven approaches to replace conventional physical models in urban air pollution forecasting. Nevertheless, it is still challenging to capture the intricate spatial and temporal patterns of air pollutant concentrations measured by heterogeneous sensors, especially for long-term prediction of the multi-variate time series data. This paper proposes a deep learning framework for longer-term forecast of air pollutants concentrations using air pollution sensing data, based on a conceptual framework of meta-graph deep learning. The key modules in the framework include meta-graph units and fusion layers, which are designed to learn temporal and spatial correlations respectively. A detailed case was formulated for forecasting air pollutants in Stockholm using air quality sensing data, meteorological data and so on. Experiments were conducted to evaluate the performance of the proposed modelling framework. The computational results show that it outperforms the baseline models and conventional deterministic dispersion models, demonstrating the potential of the framework to be deployed for the real air quality information systems in Stockholm.

Online traffic simulation that feeds from online information to simulate vehicle movement in real-time has recently seen substantial advancement in the development of intelligent transportation systems and urban traffic management. It has been a challenging problem due to three aspects: 1) The diversity of traffic patterns due to heterogeneous layouts of urban intersections; 2) The nature of complex spatiotemporal correlations; 3) The requirement of dynamically adjusting the parameters of traffic models in a real-time system. To cater to these challenges, this paper proposes an online traffic simulation framework called automated urban traffic operation simulation via meta-learning (AUTOSIM). In particular, simulation models with various intersection layouts are automatically generated using an open-source simulation tool based on static traffic geometry attributes. Through a meta-learning technique, AUTOSIM enables an automated learning process for dynamic model settings of traffic scenarios featured with different spatiotemporal correlations. Besides, AUTOSIM is capable of adapting traffic model parameters according to dynamic traffic information in real-time by using a meta-learner. Through computational experiments, we demonstrate the effectiveness of the meta-learning-based framework that is capable of providing reliable supports to real-time traffic simulation and dynamic traffic operations.

Accurate traffic forecasting plays a critical role in the construction of intelligent transportation systems. However, due to the across road-network isomorphism in the spatial dimension and the periodic drift in the temporal dimension, existing traffic forecasting methods cannot satisfy the intricate spatial-temporal characteristics well. In this article, a spatial-temporal hypergraph convolutional network for traffic forecasting (ST-HCN) is proposed to tackle the problems mentioned above. Specifically, the proposed framework applies the K-means clustering algorithm and the connection characteristics of the physical road network itself to unify the local correlation and across road-network isomorphism. Then, a dual-channel hypergraph convolution to capture high-order spatial relationships in traffic data is established. Furthermore, the proposed framework utilizes a long short-term memory network with a convolution module (ConvLSTM) to deal with the periodic drift problem. Finally, the experiments in the real world demonstrate that the proposed framework outperforms the state-of-the-art baselines.

The optimal allocation of space-time resources is essential to enhance the performance of an intersection under variant traffic demand. While most research has concentrated on optimizing signal timings via rule-based or learning-based methodologies, there is a dearth of studies on how to adjust lane configuration to further optimize space resources in this context. This is particularly pertinent given the significantly variable flow patterns and turning ratios of urban traffic flow. To address this issue, Hi-Lane, a hierarchical framework is proposed for decision support on intersection traffic control. Hi-Lane takes advantage of dynamic lanes, which have been recently implemented to facilitate better usage of space resources. Based on Hierarchical Reinforcement Learning (HRL), Hi-Lane employs two reinforcement learning agents to tackle this complex optimization problem. Initially, the lane configuration agent will recommend the direction of dynamic lanes based on long-term traffic flow patterns. Then, the signal timing agent will optimize the hyperparameters of the downstream signal control method, which finally allocates green time to the traffic light. Hi-Lane's utility and superiority are demonstrated on an intersection with simulated traffic flows, where it outperforms the models which optimize space or time resources alone.

Empirical studies on the robustness of graph neural networks (GNNs) have suggested a relation between the vulnerabilities of GNNs to adversarial attacks and the increased presence of heterophily in perturbed graphs (where edges tend to connect nodes with dissimilar features and labels). In this work, we formalize the relation between heterophily and robustness, bridging two topics previously investigated by separate lines of research. We theoretically and empirically show that for graphs exhibiting homophily (low heterophily), impactful structural attacks always lead to increased levels of heterophily, while for graph with heterophily the change in the homophily level depends on the node degrees. By leveraging these insights, we deduce that a design principle identified to significantly improve predictive performance under heterophily—separate aggregators for ego- and neighbor-embeddings—can also inherently offer increased robustness to GNNs. Our extensive empirical analysis shows that GNNs adopting this design alone can achieve significantly improved empirical and certifiable robustness compared to the best-performing unvaccinated model. Furthermore, models with this design can be readily combined with explicit defense mechanisms to yield improved robustness with up to 18.33% increase in performance under attacks compared to the best-performing vaccinated model.

Road link speed is often employed as an essential measure of traffic state in the operation of an urban traffic network. Not only real-time traffic demand but also signal timings and other local planning factors are major influential factors. This paper proposes a short-term traffic speed prediction approach, called PL-WGAN, for urban road networks, which is considered an important part of a novel parallel learning framework for traffic control and operation. The proposed method applies Wasserstein Generative Adversarial Nets (WGAN) for robust data-driven traffic modeling using a combination of generative neural network and discriminative neural network. The generative neural network models the road link features of the adjacent intersections and the control parameters of intersections using a hybrid graph block. In addition, the spatial-temporal relations are captured by stacking a graph convolutional network (GCN), a recurrent neural network (RNN), and an attention mechanism. A comprehensive computational experiment was carried out including comparing model prediction and computational performances with several state-of-the-art deep learning models. The proposed approach has been implemented and applied for predicting short-term link traffic speed in a large-scale urban road network in Hangzhou, China. The results suggest that it provides a scalable and effective traffic prediction solution for urban road networks.

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.

Strategic traffic management is crucial for combating traffic congestion at the macroscopic level. However, such a field is still relatively unexplored, particularly for microscopic control objects, such as intersections and coordinated intersection groups. This article proposes a human-in-the-loop recommendation system for strategic urban traffic management, which follows an agent-based structure. A regional agent dispatcher is defined to assign agents for operation whenever ``operation on-demand'' is required. Such a requirement is identified by a daily-dependent operational mode on strategic traffic operations at a control object level. The strategic management scheme for each control object is guided by a strategic agent (customized), which is essentially a deep recommender model with a specific architecture. By featuring the multiagent design, a customized operational scheme can be generated at the intersection level, which instructs the corresponding controller to take specific operations. The utility of the recommendation system is demonstrated via a case study using real-world traffic data. In both offline and online evaluations, the system performs consistently at traffic operational recommendations in different scenarios and has the potential to provide more reasonable traffic operational strategies than a human-operated system.

Parallel transportation management and control was proposed three decades ago as a new paradigm for conducting complex transportation operations and has led to today’s DeCAST in TransVerse platform designed and constructed according to the principle of decentralized/distributed autonomous operations and organizations. This article presents an overview of its architectures, processes, operating procedures, and major applications. The developments and applications have demonstrated clearly that parallel transportation systems are effective for networked traffic control and distributed logistical operations. The existing challenges and emerging opportunities are also addressed. A transportation foundation model based on parallel learning and federated intelligence is proposed as a potential path to the next-generation parallel intelligent transportation systems.

While most network embedding techniques model the proximity between nodes in a network, recently there has been significant interest in structural embeddings that are based on node equivalences , a notion rooted in sociology: equivalences or positions are collections of nodes that have similar roles—i.e., similar functions, ties or interactions with nodes in other positions—irrespective of their distance or reachability in the network. Unlike the proximity-based methods that are rigorously evaluated in the literature, the evaluation of structural embeddings is less mature. It relies on small synthetic or real networks with labels that are not perfectly defined, and its connection to sociological equivalences has hitherto been vague and tenuous. With new node embedding methods being developed at a breakneck pace, proper evaluation, and systematic characterization of existing approaches will be essential to progress. To fill in this gap, we set out to understand what types of equivalences structural embeddings capture. We are the first to contribute rigorous intrinsic and extrinsic evaluation methodology for structural embeddings, along with carefully-designed, diverse datasets of varying sizes. We observe a number of different evaluation variables that can lead to different results (e.g., choice of similarity measure, classifier, and label definitions). We find that degree distributions within nodes’ local neighborhoods can lead to simple yet effective baselines in their own right and guide the future development of structural embedding. We hope that our findings can influence the design of further node embedding methods and also pave the way for more comprehensive and fair evaluation of structural embedding methods.

Deep Reinforcement Learning (DRL) has proven its capability to learn optimal policies in decision-making problems by directly interacting with environments. Meanwhile, supervised learning methods also show great capability of learning from data. However, how to combine DRL with supervised learning and leverage additional knowledge and data to assist the DRL agent remains difficult. This study proposes a novel Supervised Assisted Deep Reinforcement Learning (SADRL) framework integrating deep Q-learning from dynamic demonstrations with a behavioral cloning model (DQfDD-BC). Specifically, the proposed DQfDD-BC method leverages historical demonstrations to pre-train a behavioral cloning model and consistently update it by learning the dynamically updated demonstrations. A supervised expert loss function is designed to compare actions generated by the DRL model with those obtained from the BC model to provide advantageous guidance for policy improvements. Experimental results in several OpenAI Gym environments show that the proposed approach accelerates the learning processes, and meanwhile, adapts to different performance levels of demonstrations. As illustrated in an ablation study, the dynamic demonstration and expert loss mechanisms using a BC model contribute to improving the learning convergence performance compared with the baseline models. We believe that SADRL provides an elegant framework and the proposed method can promote the integration of human experience and machine intelligence.

Online traffic simulation feeds from online information to simulate vehicle movement in real-time, which has recently seen substantial advancement in road traffic control and management. It has been a challenging problem due to three aspects: 1) the diversity of traffic patterns caused by heterogeneous layouts of urban intersections; 2) the complexity of spatiotemporal correlations; 3) the requirement of adjusting traffic model parameters in a real-time system. To cater to these challenges, this paper proposes an online traffic simulation modeling framework via a meta-learner. In particular, simulation models with various intersection layouts are automatically generated using an open-source simulation tool, SUMO, according to static traffic geometry attributes. Through a meta-learning technique, the proposed modeling framework enables an automated learning process for estimating model settings capable of adapting traffic model parameters according to dynamic traffic information in real-time. Such a process is featured with various traffic scenarios and different spatiotemporal correlations. Through computational experiments, we demonstrate that the meta-learning-based framework is able to self-adapt its effectiveness according to real-time traffic data.

This paper proposes GraphPro, a short-term link speed prediction framework for signalized urban traffic networks. Different from other traditional approaches that adopt only reactive inputs (i.e., surrounding traffic data), GraphPro also accepts proactive inputs (i.e., traffic signal timing). This allows GraphPro to predict link speed more accurately, depending on whether or not there is a contextual change in traffic signal timing. A Wasserstein generative adversarial network (WGAN) structure, including a generator (prediction model) and a discriminator, is employed to incorporate unprecedented network traffic states and ensures a high level of generalizability for the prediction model. A hybrid graph block, comprised of a reactive cell and a proactive cell, is implemented into each neural layer of the generator. In order to jointly capture spatio-temporal influences and signal contextual information on traffic links, the two cells adopt several key neural network-based components, including graph convolutional network, recurrent neural architecture, and self-attention mechanism. The double-cell structure ensures GraphPro learns from proactive input only when required. The effectiveness and efficiency of GraphPro are tested on a short-term link speed prediction task using real-world traffic data. Due to the capabilities of learning from real data distribution and generating unseen samples, GraphPro offers a more reliable and robust prediction when compared with state-of-the-art data-driven models.

This paper proposes a parallel recommendation engine, PRECOM, for traffic control operations to mitigate congestion of road traffic in the metropolitan area. The recommendation engine can provide, in real-time, effective and optimal control plans to traffic engineers, who are responsible for manually calibrating traffic signal plans especially when a road network suffers from heavy congestion due to disruptive events. With the idea of incorporating expert knowledge in the operation loop, the PRECOM system is designed to include three conceptual components: an artificial system model, a computational experiment module, and a parallel execution module. Meanwhile, three essential algorithmic steps are implemented in the recommendation engine: a candidate generator based on a graph model, a spatiotemporal ranker, and a context-aware re-ranker. The PRECOM system has been deployed in the city of Hangzhou, China, through both offline and online evaluation. The experimental results are promising, and prove that the recommendation system can provide effective support to the current human-in-the-loop control scheme in the practice of traffic control, operations, and management.

Recent studies have exposed that many graph neural networks (GNNs) are sensitive to adversarial attacks, and can suffer from performance loss if the graph structure is intentionally perturbed. A different line of research has shown that many GNN architectures implicitly assume that the underlying graph displays homophily, i.e., connected nodes are more likely to have similar features and class labels, and perform poorly if this assumption is not fulfilled. In this work, we formalize the relation between these two seemingly different issues. We theoretically show that in the standard scenario in which node features exhibit homophily, impactful structural attacks always lead to increased levels of heterophily. Then, inspired by GNN architectures that target heterophily, we present two designs -- (i) separate aggregators for ego- and neighbor-embeddings, and (ii) a reduced scope of aggregation -- that can significantly improve the robustness of GNNs. Our extensive empirical evaluations show that GNNs featuring merely these two designs can achieve significantly improved robustness compared to the best-performing unvaccinated model with 24.99% gain in average performance under targeted attacks, while having smaller computational overhead than existing defense mechanisms. Furthermore, these designs can be readily combined with explicit defense mechanisms to yield state-of-the-art robustness with up to 18.33% increase in performance under attacks compared to the best-performing vaccinated model.

Although Deep Reinforcement Learning (DRL) has proven its capability to learn optimal policies by directly interacting with simulation environments, how to combine DRL with supervised learning and leverage additional knowledge to assist the DRL agent effectively still remains difficult. This study proposes a novel approach integrating deep Q learning from dynamic demonstrations with a behavioral cloning model (DQfDD-BC), which includes a supervised learning technique of instructing a DRL model to enhance its performance. Specifically, the DQfDD-BC model leverages historical demonstrations to pre-train a supervised BC model and consistently update it by learning the dynamically updated demonstrations. Then the DQfDD-BC model manages the sample complexity by exploiting both the historical and generated demonstrations. An expert loss function is designed to compare actions generated by the DRL model with those obtained from the BC model to provide advantageous guidance for policy improvements. Experimental results in several OpenAI Gym environments show that the proposed approach adapts to different performance levels of demonstrations, and meanwhile, accelerates the learning processes. As illustrated in an ablation study, the dynamic demonstration and expert loss mechanisms with the utilization of a BC model contribute to improving the learning convergence performance compared with the origin DQfD model.

A paradigm shift towards agile and adaptive traffic signal control empowered with the massive growth of Big Data and Internet of Things (IoT) technologies is emerging rapidly for Intelligent Transportation Systems. Generally, an adaptive signal control system fine-tunes signal timing parameters based on pre-defined control hyperparameters using instantaneous traffic detection information. Once traffic pattern changes, those hyperparameters (e.g., maximum and minimum green times) need to be adjusted according to the evolution of traffic dynamics over a very short-term period. Such adjustment processes are usually conducted by professional and experienced traffic engineers. Here we present a human-in-the-loop parallel learning framework and its utilization in an end-to-end recommendation system that mimics and enhances professional signal control engineers' behaviors. The system has been deployed into a real-world application for an extended period in Hangzhou, China, where signal control hyperparameters are recommended based on large-scale multidimensional traffic datasets. Experimental evaluations demonstrate significant improvements in traffic efficiency through the use of our signal recommendation system.