Graph Attention Informer for Long-Term Traffic Flow Prediction under the Impact of Sports Events.

Traffic flow prediction is one of the challenges in the development of an Intelligent Transportation System (ITS). Accurate traffic flow prediction helps to alleviate urban traffic congestion and improve urban traffic efficiency, which is crucial for promoting the synergistic development of smart transportation and smart cities. With the development of deep learning, many deep neural networks have been proposed to address this problem. However, due to the complexity of traffic maps and external factors, such as sports events, these models cannot perform well in long-term prediction. In order to enhance the accuracy and robustness of the model on long-term time series prediction, a Graph Attention Informer (GAT-Informer) structure is proposed by combining the graph attention layer and informer layer to capture the intrinsic features and external factors in spatial-temporal correlation. The external factors are represented as sports events impact factors. The GAT-Informer model was tested on real-world data collected in London, and the experimental results showed that our model has better performance in long-term traffic flow prediction compared to other baseline models.

A Dual-Stream Cross AGFormer-GPT Network for Traffic Flow Prediction Based on Large-Scale Road Sensor Data.

Traffic flow prediction can provide important reference data for managers to maintain traffic order, and can also be based on personal travel plans for optimal route selection. On account of the development of sensors and data collection technology, large-scale road network historical data can be effectively used, but their high non-linearity makes it meaningful to establish effective prediction models. In this regard, this paper proposes a dual-stream cross AGFormer-GPT network with prompt engineering for traffic flow prediction, which integrates traffic occupancy and speed as two prompts into traffic flow in the form of cross-attention, and uniquely mines spatial correlation and temporal correlation information through the dual-stream cross structure, effectively combining the advantages of the adaptive graph neural network and large language model to improve prediction accuracy. The experimental results on two PeMS road network data sets have verified that the model has improved by about 1.2% in traffic prediction accuracy under different road networks.

Parking Lot Traffic Prediction Based on Fusion of Multifaceted Spatio-Temporal Features.

With the rapid growth of population and vehicles, issues such as traffic congestion are becoming increasingly apparent. Parking guidance and information (PGI) systems are becoming more critical, with one of the most important tasks being the prediction of traffic flow in parking lots. Predicting parking traffic can effectively improve parking efficiency and alleviate traffic congestion, traffic accidents, and other problems. However, due to the complex characteristics of parking spatio-temporal data, high levels of noise, and the intricate influence of external factors, there are three challenges to predicting parking traffic in a city effectively: (1) how to better model the nonlinear, asymmetric, and complex spatial relationships among parking lots; (2) how to model the temporal autocorrelation of parking flow more accurately for each parking lot, whether periodic or aperiodic; and (3) how to model the correlation between external influences, such as holiday weekends, POIs (points of interest), and weather factors. In this context, this paper proposes a parking lot traffic prediction model based on the fusion of multifaceted spatio-temporal features (MFF-STGCN). The model consists of a feature embedding module, a spatio-temporal attention mechanism module, and a spatio-temporal convolution module. The feature embedding module embeds external features such as weekend holidays, geographic POIs, and weather features into the time series, the spatio-temporal attention mechanism module captures the dynamic spatio-temporal correlation of parking traffic, and the spatio-temporal convolution module captures the spatio-temporal features by using graph convolution and gated recursion units. Finally, the outputs of adjacent time series, daily series, and weekly series are weighted and fused to obtain the final prediction results, thus predicting the parking lot traffic flow more accurately and effectively. Results on real datasets demonstrate that the proposed model enhances prediction performance.

A Dynamic Traffic Light Control Algorithm to Mitigate Traffic Congestion in Metropolitan Areas.

This paper proposes a convolutional neural network (CNN) model of the signal distribution control algorithm (SDCA) to maximize the dynamic vehicular traffic signal flow for each junction phase. The aim of the proposed algorithm is to determine the reward value and new state. It deconstructs the routing components of the current multi-directional queuing system (MDQS) architecture to identify optimal policies for every traffic scenario. Initially, the state value is divided into a function value and a parameter value. Combining these two scenarios updates the resulting optimized state value. Ultimately, an analogous criterion is developed for the current dataset. Next, the error or loss value for the present scenario is computed. Furthermore, utilizing the Deep Q-learning methodology with a quad agent enhances previous study discoveries. The recommended method outperforms all other traditional approaches in effectively optimizing traffic signal timing.

Random Traffic Flow Simulation of Heavy Vehicles Based on R-Vine Copula Model and Improved Latin Hypercube Sampling Method.

The rationality of heavy vehicle models is crucial to the structural safety assessment of bridges. To establish a realistic heavy vehicle traffic flow model, this study proposes a heavy vehicle random traffic flow simulation method that fully considers the vehicle weight correlation based on the measured weigh-in-motion data. First, a probability model of the key parameters in the actual traffic flow is established. Then, a random traffic flow simulation of heavy vehicles is realized using the R-vine Copula model and improved Latin hypercube sampling (LHS) method. Finally, the load effect is calculated using a calculation example to explore the necessity of considering the vehicle weight correlation. The results indicate that the vehicle weight of each model is significantly correlated. Compared to the Monte Carlo method, the improved LHS method better considers the correlation between high-dimensional variables. Furthermore, considering the vehicle weight correlation using the R-vine Copula model, the random traffic flow generated by the Monte Carlo sampling method ignores the correlation between parameters, leading to a weaker load effect. Therefore, the improved LHS method is preferred.

MD-GCN: A Multi-Scale Temporal Dual Graph Convolution Network for Traffic Flow Prediction.

The spatial-temporal prediction of traffic flow is very important for traffic management and planning. The most difficult challenges of traffic flow prediction are the temporal feature extraction and the spatial correlation extraction of nodes. Due to the complex spatial correlation between different roads and the dynamic trend of time patterns, traditional forecasting methods still have limitations in obtaining spatial-temporal correlation, which makes it difficult to extract more valid information. In order to improve the accuracy of the forecasting, this paper proposes a multi-scale temporal dual graph convolution network for traffic flow prediction (MD-GCN). Firstly, we propose a gated temporal convolution based on a channel attention and inception structure to extract multi-scale temporal dependence. Then, aiming at the complexity of the traffic spatial structure, we develop a dual graph convolution module including the graph sampling and aggregation submodule (GraphSAGE) and the mix-hop propagation graph convolution submodule (MGCN) to extract the local correlation and global correlation between neighbor nodes. Finally, extensive experiments are carried out on several public traffic datasets, and the experimental results show that our proposed algorithm outperforms the existing methods.

Research on Traffic Flow Prediction at Intersections Based on DT-TCN-Attention.

Traditional nonintelligent signal control systems are typically used in road traffic signal systems, which cannot provide optimal guidance and have low traffic efficiency during rush hour. This study proposes a traffic signal phase dynamic timing optimization strategy based on a time convolution network and attention mechanism to improve traffic efficiency at intersections. The corresponding optimization was performed after predicting traffic conditions with different impacts using the digital twinning technique. This method uses a time-convolution network to extract the cross-time nonlinear characteristics of traffic data at road intersections. An attention mechanism was introduced to capture the relationship between the importance distribution and duration of the historical time series to predict the traffic flow at an intersection. The interpretability and prediction accuracy of the model was effectively improved. The model was tested using traffic flow data from a signalized intersection in Shangrao, Jiangxi Province, China. The experimental results indicate that the model generated by training has a strong learning ability for the temporal characteristics of traffic flow. The model has high prediction accuracy, good optimization results, and wide application prospects in different scenarios.

Unlocking Mutual Gains-An Experimental Study on Collaborative Autonomous Driving in Urban Environment.

Convoy driving, a specialized form of collaborative autonomous driving, offers a promising solution to the multifaceted challenges that transportation systems face, including traffic congestion, pollutant emissions, and the coexistence of connected autonomous vehicles (CAVs) and human-driven vehicles on the road, resulting in mixed traffic flow. While extensive research has focused on the collective societal benefits of convoy driving, such as safety and comfort, one critical aspect that has been overlooked is the willingness of individual vehicles to participate in convoy formations. While the collective benefits are evident, individual vehicles may not readily embrace this paradigm shift without explicit tangible benefits and incentives to motivate them. Moreover, the objective of convoy driving is not solely to deliver societal benefits but also to provide incentives and reduce costs at the individual level. Therefore, this research bridges this gap by designing and modeling the societal benefits, including traffic flow optimization and pollutant emissions, and individual-level incentives necessary to promote convoy driving. We model a fundamental diagram of mixed traffic flow, considering various factors such as CAV penetration rates, coalition intensity, and coalition sizes to investigate their relationships and their impact on traffic flow. Furthermore, we model the collaborative convoy driving problem using the coalitional game framework and propose a novel utility function encompassing incentives like car insurance discounts, traffic fine reductions, and toll discounts to encourage vehicle participation in convoys. Our experimental findings emphasize the need to strike a balance between CAV penetration rate, coalition intensity, size, and speed to realize the benefits of convoy driving at both collective and individual levels. This research aims to align the interests of road authorities seeking sustainable transportation systems and individual vehicle owners desiring tangible benefits, envisioning a future where convoy driving becomes a mutually beneficial solution.

A CNN-LSTM Car-Following Model Considering Generalization Ability.

To explore the potential relationship between the leading vehicle and the following vehicle during car-following, we proposed a novel car-following model combining a convolutional neural network (CNN) with a long short-term memory (LSTM) network. Firstly, 400 car-following periods were extracted from the natural driving database and the OpenACC car-following experiment database. Then, we developed a CNN-LSTM car-following model, and the CNN is employed to analyze the potential relationship between the vehicle's dynamic parameters and to extract the features of car-following behavior to generate the feature vector. The LSTM network is adopted to save the feature vector and predict the speed of the following vehicle. Finally, the CNN-LSTM model is trained and tested with the extracted car-following trajectories data and compared with the classical car-following models (LSTM model, intelligent driver model). The results show that the accuracy and the ability to learn the heterogeneity of the proposed model are better than the other two. Furthermore, the CNN-LSTM model can accurately reproduce the hysteresis phenomenon of congested traffic flow and apply to heterogeneous traffic flow mixed with adaptive cruise control vehicles on the freeway, which indicates that it has strong generalization ability.

Stochastic Propagation of Fatigue Cracks in Welded Joints of Steel Bridge Decks under Simulated Traffic Loading.

The fatigue cracking of orthotropic steel bridge decks (OSDs) is a difficult problem that hinders the development of steel structures. The most important reasons for the occurrence of fatigue cracking are steadily growing traffic loads and unavoidable truck overloading. Stochastic traffic loading leads to the random propagation behavior of fatigue cracks, which increases the difficulty of the fatigue life evaluations of OSDs. This study developed a computational framework for the fatigue crack propagation of OSDs under stochastic traffic loads based on traffic data and finite element methods. Stochastic traffic load models were established based on site-specific, weigh-in-motion measurements to simulate fatigue stress spectra of welded joints. The influence of the transverse loading positions of the wheel tracks on the stress intensity factor of the crack tip was investigated. The random propagation paths of the crack under stochastic traffic loads were evaluated. Both ascending and descending load spectra were considered in the traffic loading pattern. The numerical results indicated that the maximum value of KI was 568.18 (MPa·mm1/2) under the most critical transversal condition of the wheel load. However, the maximum value decreased by 66.4% under the condition of transversal moving by 450 mm. In addition, the propagation angle of the crack tip increased from 0.24° to 0.34°-an increase ratio of 42%. Under the three stochastic load spectra and the simulated wheel loading distributions, the crack propagation range was almost limited to within 10 mm. The migration effect was the most obvious under the descending load spectrum. The research results of this study can provide theoretical and technical support for the fatigue and fatigue reliability evaluation of existing steel bridge decks.

Multi-Agent Reinforcement Learning for Traffic Flow Management of Autonomous Vehicles.

Intelligent traffic management systems have become one of the main applications of Intelligent Transportation Systems (ITS). There is a growing interest in Reinforcement Learning (RL) based control methods in ITS applications such as autonomous driving and traffic management solutions. Deep learning helps in approximating substantially complex nonlinear functions from complicated data sets and tackling complex control issues. In this paper, we propose an approach based on Multi-Agent Reinforcement Learning (MARL) and smart routing to improve the flow of autonomous vehicles on road networks. We evaluate Multi-Agent Advantage Actor-Critic (MA2C) and Independent Advantage Actor-Critical (IA2C), recently suggested Multi-Agent Reinforcement Learning techniques with smart routing for traffic signal optimization to determine its potential. We investigate the framework offered by non-Markov decision processes, enabling a more in-depth understanding of the algorithms. We conduct a critical analysis to observe the robustness and effectiveness of the method. The method's efficacy and reliability are demonstrated by simulations using SUMO, a software modeling tool for traffic simulations. We used a road network that contains seven intersections. Our findings show that MA2C, when trained on pseudo-random vehicle flows, is a viable methodology that outperforms competing techniques.

A Dynamic Lane-Changing Driving Strategy for CAV in Diverging Areas Based on MPC System.

Freeway-diverging areas are prone to low traffic efficiency, congestion, and frequent accidents. Because of the fluctuation of the surrounding traffic flow distribution, the individual decision-making of vehicles in diverging areas is typically unable to plan a departure trajectory that balances safety and efficiency well. Consequently, it is critical that vehicles in freeway-diverging regions develop a lane-changing driving strategy that strives to improve both the safety and efficiency of divergence areas. For CAV leaving the diverging area, this study suggested a full-time horizon optimum solution. Since it is a dynamic strategy, an MPC system based on rolling time horizon optimization was constructed as the primary algorithm of the strategy. A simulation experiment was created to verify the viability of the proposed methodology based on a mixed-flow environment. The results show that, in comparison with the feasible strategies exiting to off-ramp, the proposed strategy can take over 60% reduction in lost time traveling through a diverging area under the premise of safety and comfort without playing a negative impact on the surrounding traffic flow. Thus, the MPC system designed for the subject vehicle is capable of performing an optimal driving strategy in diverging areas within the full-time and space horizon.

A Collaborative Merging Method for Connected and Automated Vehicle Platoons in a Freeway Merging Area with Considerations for Safety and Efficiency.

To solve the problems of congestion and accident risk when multiple vehicles merge into the merging area of a freeway, a platoon split collaborative merging (PSCM) method was proposed for an on-ramp connected and automated vehicle (CAV) platoon under a mixed traffic environment composed of human-driving vehicles (HDV) and CAVs. The PSCM method mainly includes two parts: merging vehicle motion control and merging effect evaluation. Firstly, the collision avoidance constraints of merging vehicles were analyzed, and on this basis, a following-merging motion rule was proposed. Then, considering the feasibility of and constraints on the stability of traffic flow during merging, a performance measurement function with safety and merging efficiency as optimization objectives was established to screen for the optimal splitting strategy. Simulation experiments under traffic demand of 1500 pcu/h/lane and CAV ratios of 30%, 50%, and 70% were conducted respectively. It was shown that under the 50% CAV ratio, the average travel time of the on-ramp CAV platoon was reduced by 50.7% under the optimal platoon split strategy compared with the no-split control strategy. In addition, the average travel time of main road vehicles was reduced by 27.9%. Thus, the proposed PSCM method is suitable for the merging control of on-ramp CAV platoons under the condition of heavy main road traffic demand.

Attention-Based Spatial-Temporal Convolution Gated Recurrent Unit for Traffic Flow Forecasting.

Accurate traffic flow forecasting is very important for urban planning and traffic management. However, this is a huge challenge due to the complex spatial-temporal relationships. Although the existing methods have researched spatial-temporal relationships, they neglect the long periodic aspects of traffic flow data, and thus cannot attain a satisfactory result. In this paper, we propose a novel model Attention-Based Spatial-Temporal Convolution Gated Recurrent Unit (ASTCG) to solve the traffic flow forecasting problem. ASTCG has two core components: the multi-input module and the STA-ConvGru module. Based on the cyclical nature of traffic flow data, the data input to the multi-input module are divided into three parts, near-neighbor data, daily-periodic data, and weekly-periodic data, thus enabling the model to better capture the time dependence. The STA-ConvGru module, formed by CNN, GRU, and attention mechanism, can capture both temporal and spatial dependencies of traffic flow. We evaluate our proposed model using real-world datasets and experiments show that the ASTCG model outperforms the state-of-the-art model.

A Physics-Informed Generative Car-Following Model for Connected Autonomous Vehicles.

This paper proposes a novel hybrid car-following model: the physics-informed conditional generative adversarial network (PICGAN), designed to enhance multi-step car-following modeling in mixed traffic flow scenarios. This hybrid model leverages the strengths of both physics-based and deep-learning-based models. By taking advantage of the inherent structure of GAN, the PICGAN eliminates the need for an explicit weighting parameter typically used in the combination of traditional physics-based and data-driven models. The effectiveness of the proposed model is substantiated through case studies using the NGSIM I-80 dataset. These studies demonstrate the model's superior trajectory reproduction, suggesting its potential as a strong contender to replace conventional models in trajectory prediction tasks. Furthermore, the deployment of PICGAN significantly enhances the stability and efficiency in mixed traffic flow environments. Given its reliable and stable results, the PICGAN framework contributes substantially to the development of efficient longitudinal control strategies for connected autonomous vehicles (CAVs) in real-world mixed traffic conditions.

Forecasting metro rail transit passenger flow with multiple-attention deep neural networks and surrounding vehicle detection devices.

In the rapid development of public transportation led, the traffic flow prediction has become one of the most crucial issues, especially estimating the number of passengers using the Mass Rapid Transit (MRT) system. In general, predicting the passenger flow of traffic is a time-series problem that requires external information to improve accuracy. Because many MRT passengers take cars or buses to MRT stations, this study used external information from vehicle detection (VD) devices to improve the prediction of passenger flow. This study proposed a deep learning architecture, called a multiple-attention deep neural network (MADNN) model, based on historical MRT passenger flow and the flow from surrounding VD devices that estimates the weights of the vehicle detection devices. The model consists of (1) an MRT attention layer (MRT-AL) that generate hidden features for MRT stations, (2) a surrounding VD (SVD) attention layer (SVD-AL) that generate hidden features for SVD devices, and (3) an MRT-SVD attention layer (MRT-SVD-AL) that generate attention weights for each VD device in an MRT station. The results of the investigation indicated that the MADNN model outperformed the models without multiple-attention mechanisms in predicting the passenger flow of MRT traffic.

Mobility Dynamics amid COVID-19 with a Case Study in Tennessee.

The year 2020 has marked the spread of a global pandemic, COVID-19, challenging many aspects of our daily lives. Different organizations have been involved in controlling this outbreak. The social distancing intervention is deemed to be the most effective policy in reducing face-to-face contact and slowing down the rate of infections. Stay-at-home and shelter-in-place orders have been implemented in different states and cities, affecting daily traffic patterns. Social distancing interventions and fear of the disease resulted in a traffic decline in cities and counties. However, after stay-at-home orders ended and some public places reopened, traffic gradually started to revert to pre-pandemic levels. It can be shown that counties have diverse patterns in the decline and recovery phases. This study analyzes county-level mobility change after the pandemic, explores the contributing factors, and identifies possible spatial heterogeneity. To this end, 95 counties in Tennessee have been selected as the study area to perform geographically weighted regressions (GWR) models. The results show that density on non-freeway roads, median household income, percent of unemployment, population density, percent of people over age 65, percent of people under age 18, percent of work from home, and mean time to work are significantly correlated with vehicle miles traveled change magnitude in both decline and recovery phases. Also, the GWR estimation captures the spatial heterogeneity and local variation in coefficients among counties. Finally, the results imply that the recovery phase could be estimated depending on the identified spatial attributes. The proposed model can help agencies and researchers estimate and manage decline and recovery based on spatial factors in similar events in the future.

Convolutional Long-Short Term Memory Network with Multi-Head Attention Mechanism for Traffic Flow Prediction.

Accurate predictive modeling of traffic flow is critically important as it allows transportation users to make wise decisions to circumvent traffic congestion regions. The advanced development of sensing technology makes big data more affordable and accessible, meaning that data-driven methods have been increasingly adopted for traffic flow prediction. Although numerous data-driven methods have been introduced for traffic flow predictions, existing data-driven methods cannot consider the correlation of the extracted high-dimensional features and cannot use the most relevant part of the traffic flow data to make predictions. To address these issues, this work proposes a decoder convolutional LSTM network, where the convolutional operation is used to consider the correlation of the high-dimensional features, and the LSTM network is used to consider the temporal correlation of traffic flow data. Moreover, the multi-head attention mechanism is introduced to use the most relevant portion of the traffic data to make predictions so that the prediction performance can be improved. A traffic flow dataset collected from the Caltrans Performance Measurement System (PeMS) database is used to demonstrate the effectiveness of the proposed method.

GLSNN Network: A Multi-Scale Spatiotemporal Prediction Model for Urban Traffic Flow.

Traffic flow prediction is a key issue in intelligent transportation systems. The growing trend in data disclosure has created more potential sources for the input for predictive models, posing new challenges to the prediction of traffic flow in the era of big data. In this study, the prediction of urban traffic flow was regarded as a spatiotemporal prediction problem, focusing on the traffic speed. A Graph LSTM (Long Short-Term Memory) Spatiotemporal Neural Network (GLSNN) model was constructed to perform a multi-scale spatiotemporal fusion prediction based on the multi-source input data. The GLSNN model consists of three parts: MS-LSTM, LZ-GCN, and LSTM-GRU. We used the MS-LSTM module to scale the traffic timing data, and then used the LZ-GCN network and the LSTM-GRU network to capture both the time and space dependencies. The model was tested on a real traffic dataset, and the experiment results verified the superior performance of the GLSNN model on both a high-precision and multi-scale prediction of urban traffic flow.

Pedestrian Traffic Light Control with Crosswalk FMCW Radar and Group Tracking Algorithm.

The increased mobility requirements of modern lifestyles put more stress on existing traffic infrastructure, which causes reduced traffic flow, especially in peak traffic hours. This calls for new and advanced solutions in traffic flow regulation and management. One approach towards optimisation is a transition from static to dynamic traffic light intervals, especially in spots where pedestrian crossing cause stops in road traffic flow. In this paper, we propose a smart pedestrian traffic light triggering mechanism that uses a Frequency-modulated continuous-wave (FMCW) radar for pedestrian detection. Compared to, for example, camera-surveillance systems, radars have advantages in the ability to reliably detect pedestrians in low-visibility conditions and in maintaining privacy. Objects within a radar's detection range are represented in a point cloud structure, in which pedestrians form clusters where they lose all identifiable features. Pedestrian detection and tracking are completed with a group tracking (GTRACK) algorithm that we modified to run on an external processor and not integrated into the used FMCW radar itself. The proposed prototype has been tested in multiple scenarios, where we focused on removing the call button from a conventional pedestrian traffic light. The prototype responded correctly in practically all cases by triggering the change in traffic signalization only when pedestrians were standing in the pavement area directly in front of the zebra crossing.