The increasingly complicated urban traffic patterns lead traffic signal control to a new trend of higher flexibility and quicker response, which becomes possible with advances in both sensor technology and artificial intelligence. Though in its early stage, existing intelligent signal controllers equipped with reinforcement learning (RL)-based feature extractor and large language model (LLM)-driven scenario understanding and decision support already demonstrate powerful data digesting ability. This study thus proposes a smart traffic light control system integrating a vision-based perception tool to extract traffic state from real-time snapshot image of the intersection, and an LLM agent controller for signal phase switching upon scenario analysis. An indicator describing the urgency for green time at phase level is defined to abstract the contextual information regarding the competition of multiple approaching traffic flows, which augments the LLM with domain-specific logical reasoning for signal control action generation, aimed at assigning green time to the flows with the most compelling needs. With a RL-based controller providing initial control decision as backup, the proposed method is able to handle both pre-trained and out-of-distribution scenarios through real-time traffic state diagnosis and knowledgeable reasoning. Simulation evaluation on different intersection layouts and vehicle compositions is conducted with horizontal comparison of five benchmarks. A decrease in average waiting time was realized by more than 5 % under normal traffic scenario and 20 % under emergency vehicle scenario, respectively. Further, comprehensive analysis was conducted to explore the applicability of the proposed method and feasibility for real-world application in unmanned aerial vehicle (UAV)-based intelligent traffic management.

The urban signalized road network, characterized by its dynamic and complex nature due to frequent signal control adjustments and unpredictable demand fluctuations, presents significant challenges for predicting lane-level traffic flow. This study introduces the innovative MGCN-TAMA model, which addresses these challenges by integrating multi-graph convolutional networks with a temporal-aware multi-head attention mechanism. The proposed model employs three types of adjacency matrices-a geographical matrix, a signal matrix, and an attention matrix-to capture the complex spatial dependencies among various traffic approaches. Additionally, the model utilizes temporal-aware multi-head attention to discern the nonlinear correlations in traffic variations over time. Tested on a real-world dataset from Tongxiang City, the MGCN-TAMA model significantly outperforms traditional models. Notably, in the first 30-minute prediction interval, our model achieves the lowest Mean Absolute Error, with 2.5649 vehicles per 5-minute span. These results underscore the effectiveness of combining graph-based methods with advanced attention mechanisms to enhance the accuracy of predicting lane-level traffic volumes in urban networks.

Cycle-based arrival profiles can describe temporal demand distribution within a signal cycle for signalized intersections, which can be used to calculate indicators such as traffic volume, queue length, and facilitate fine-grained signal control. However, few studies address cycle-level arrival profile estimation based on connected vehicles (CVs). Besides, studies addressing privacy issues for cross-company collaboration in traffic management are still in their infancy. To fill these research gaps, this study proposes a data-driven method for privacy-preserving cycle-based arrival profile estimation using cross-company CV data. The cyclic arrival curve is discretized as an arrival rate vector whose elements are calculated using sampled CV trajectories, thus transforming the arrival profile estimation into a matrix completion problem. Considering cross-company collaboration, a privacy-preserving technique, secure multi-party computation, is used to encrypt initial arrival rate matrices of multiple companies. In particular, a perturbation approach is combined to enhance protection against inference attacks with prior knowledge of the matrix construction process. Then, matrix completion is realized through a singular value thresholding (SVT) algorithm, meanwhile achieving denoising. Empirical evaluation shows that the estimation accuracy of traffic volume and queue length derived from the proposed arrival profile estimation method can reach 87.6% and 78.4%, respectively, meanwhile protecting the privacy of multiple participating companies and outperforming existing methods. Simulation evaluation on a large-scale network further demonstrates the reliability of the proposed method considering ever-changing demand scenarios. A comprehensive sensitivity analyses exhibit its robustness to CV sample size, number of participating parties and data disparity, showing wide popularization and application prospects.

Multi-section license plate recognition (LPR) data has emerged as a valuable source for lane-based queue length estimation, providing both input–output information and sampled travel times. However, existing studies often rely on restrictive assumptions such as the first-in-first-out (FIFO) rule and uniform arrival processes, which fail to capture the complexity of multi-lane scenarios, particularly regarding overtaking behaviors and traffic flow variations. To address this issue, we propose a probabilistic approach to derive the stochastic queue length by constructing a conditional probability model of no-delay arrival time (NAT), i.e., the arrival time of vehicles without experiencing any delay, based on multi-section LPR data. First, the NAT conditions for all vehicles are established based on upstream and downstream vehicle departure times and sequences. To reduce the computational dimensionality and complexity, a dynamic programming (DP)-based algorithm is developed for vehicle group partitioning based on potential interactions between vehicles. Then, the conditional probability of NATs of each vehicle group is derived and a Markov Chain Monte Carlo (MCMC) sampling method is employed for calculation. Subsequently, the stochastic queue profile and maximum queue length for each cycle can be derived based on the NATs of vehicles. Eventually, we extend our approach to multi-lane scenarios, where the problem can be converted to a weighted general exact coverage problem and solved by a backtracking algorithm with heuristics. Empirical and simulation experiments demonstrate that our approach outperforms the baseline method, demonstrating significant improvements in accuracy and robustness across various traffic conditions, including different V/C ratios, matching rates, miss detection rates, and FIFO violation rates. The estimated queue profiles demonstrate practical value for offset optimization in traffic signal control, achieving a 6.63% delay reduction compared to the conventional method.

Automatic guided vehicle (AGV) fleet management always plays a significant role in smart manufacturing, which is widely studied as a representative nondeterministic polynomial-hard combinatorial optimization problem. With more smart factories featuring specialization in production line and human-robot interaction, AGVs are commonly bound with specific tracks, loading and unloading stations, which makes the current routing algorithms fail to play their path searching ability in complicated network topology. Thus, an integrated timetable optimization and AGV dispatching (TOAD) model is proposed aimed at such case, shifting the emphasis of routing to station selection and route selection from the perspective of timetable designing, while still considering the mixed directivity of layout, conflict avoidance, AGV availability and charging requirements. Targeted at makespan minimization, an improved genetic algorithm (GA) is used for solution with a heuristic operator to seek a better solution within shorter time. The proposed method is evaluated using an empirical factory case study with field data as input, with a comparison with the exact algorithm and standard GA. Results show that a smaller makespan and a shorter computation time can be obtained by the proposed TOAD model in large-scale scenarios, demonstrating a promising application prospect.

Among real-time traffic control methods, max-pressure (MP) control stands out due to its simplicity, decentralized nature, and robust theoretical foundation. Besides, advancements in connected vehicle (CV) technology have motivated a significant amount of research into traffic signal control based on CVs. Nevertheless, few studies have been dedicated to MP control in partially CV environments and meanwhile consider multi-modal traffic flows. To fill this research gap, this study proposes CV-based multi-modal MP control (CV-MMP), which calculates the pressure based on travel time information of CVs weighted by vehicle occupancy. Therefore, a hierarchical multi-modal traffic signal priority controller is achieved in a soft manner. Besides, adapting to the requirements of practical applications, CV-MMP is extended to fuse detector data and consider phase switching lost time and cyclic phase sequence. The evaluation results based on a toy network simulation demonstrate that CV-MMP can significantly reduce transit delay with a small increase in private vehicle delay, resulting in a significant reduction in average person delay. In addition, approximately 75% of CBs pass through the network without experiencing delays due to stopping. Therefore, our method can achieve effective transit signal priority and even transit signal coordination under single transit requests.

Network signal coordination control is a crucial means to improve the traffic operation efficiency of the overall roadway network. Accurate identification of critical paths does play an important role in determining the scope of network coordination control. Therefore, this paper proposed the definition of critical path from the perspective of traffic control and management. Under the detection environment of connected vehicle (CV), a comprehensive quantitative indicator system for path criticality evaluation from three aspects, supply side, demand side and operation side, which are arranged in the form of a tower structure. A critical path identification method (CPIM) was then proposed based on the analytic hierarchy process (AHP) theory, which was hereinafter referred to as AHP-CPIM. In order to evaluate the feasibility and effectiveness of the proposed method, a case study set in an urban network in Tongxiang, Zhejiang Province in China, is conducted through simulation models built through VISSIM and Synchro. Two scenarios were set, one is coordination control based on the coordination subarea obtained from Synchro (namely without critical path identification), and another one is coordination control with critical paths obtained from AHP-CPIM. Results showed that, compared with the control of Synchro and Multiband method under the scenario of coordination control without critical path identification, network signal coordination control optimization based on AHP-CPIM improved about 37.9% and 35.9% in average delay, respectively, justifying the effectiveness of CV-driven critical path identification for network signal coordination control.

Complex traffic scenes greatly challenge the road safety of automated vehicles (AVs). Recent work only provides an independent perspective from the fundamental modules. This paper integrates the decision-making and path-planning modules to ensure the autonomous driving performance in the high-speed cruising scenario. First, to guarantee deep exploration of the reinforcement learning method, a Bootstrapped deep-Q-Network (BDQN) is proposed to address the adaptive decision-making of AVs. Then, quantifying the multi-performance requirements of AVs under high-speed cruising can be complex. We employ an inverse reinforcement learning (IRL) approach to learn path-planning ability from skilled drivers, generating a reference path for executing lane changes. The simulation results demonstrate the proposed framework can ensure the autonomous cruising performance with safety guarantees.