In intelligent and connected transportation systems, surveillance cameras serve as critical perception devices for real-time monitoring of traffic and lane conditions. However, haze causes light scattering and contrast reduction, degrading high-frequency details (edges, textures) and semantic information (traffic participants) in camera images. Existing methods face two major challenges in handling hazy images: (1) traditional CNNs struggle to recover high-frequency details and structures lost in haze; (2) existing methods lack explicit modeling of spatially non-uniform haze. To resolve these problems, this paper uses cascaded wavelet transform convolution to reconstruct high-frequency details. Subsequently, this paper designs a dual residual attention mechanism that emphasizes crucial semantic regions and high-frequency details in both channel and spatial dimensions. Lastly, global dehazing is achieved by Swin Transformer–based image modeling, where hierarchical window attention and residual connections effectively capture multi-scale features and long-range dependencies in traffic scenes. Experimental results show that the proposed non-uniform haze removal model improves the robustness of IoT-enabled traffic monitoring. It achieves average improvements of at least 10.45%, 4.80%, and 10.77% in PSNR, SSIM, and VSNR, respectively, and reduces LPIPS by no less than 20.44%.

Traffic oscillations present significant challenges to road transportation systems, resulting in reduced fuel efficiency, heightened crash risks, and severe congestion. Recently emerging Augmented Intelligence of Things (AIoT) technology holds promise for enhancing traffic flow through vehicle-road cooperation. A representative application involves using deep reinforcement learning (DRL) techniques to control connected autonomous vehicle (CAV) platoons to alleviate traffic oscillations. However, uncertainties in human-driven vehicles (HDVs) driving behavior and the random distribution of CAVs make it challenging to achieve effective traffic oscillation alleviation in the Internet of Things environment. Existing DRL-based mixed vehicular platoon control strategies underutilize downstream traffic data, impairing CAVs’ ability to predict and mitigate traffic oscillations, leading to inefficient speed adjustments and discomfort. This article proposes a dense communication cooperative RL policy for mixed vehicular platoons to address these challenges. It employs a parameter-sharing structure and a dense information flow topology, enabling CAVs to proactively respond to traffic oscillations while accommodating arbitrary vehicle distributions and communication failures. Experimental results demonstrate superior performance of the proposed strategy in driving efficiency, comfort, and safety, particularly in scenarios involving multivehicle cut-ins or cut-outs and communication failures.

Ensuring the quality of trajectories is of utmost importance in traffic flow analysis. Traditional approaches rely on reconstructing nearly complete trajectories and subsequently denoising them. However, low detection rates often pose challenges and result in failed trajectory construction. To overcome this issue, this paper presents a trajectory concatenation method that combines NS Transformer prediction and Siamese-VGG16 similarity confirmation, specifically designed to address low detection rates. The employed transformer model can withstand missing values, efficiently extracting internal associations among multiple traffic parameters in conditions of sparse data. Furthermore, a lightweight image feature similarity verification step is integrated after trajectory prediction to find the most similar target to the image in the predicted spatiotemporal domain. Additionally, a lightweight image feature similarity verification step is integrated after trajectory prediction to identify the most similar targets within the predicted spatiotemporal domain. Experimental results demonstrate the efficacy of the proposed method, successfully connecting over 80% of fragmented tracks and yielding significant maintenance of MOTA above 0.74 under low detection accuracy.

Deep reinforcement learning (DRL) algorithms often face challenges in achieving stability and efficiency due to significant policy gradient variance and inaccurate reward function estimation in complex scenarios. This study addresses these issues in the context of multi-objective car-following control tasks with time lag in traffic oscillations. We propose an expert demonstration reinforcement learning (EDRL) approach that aims to stabilize training, accelerate learning, and enhance car-following performance. The key idea is to leverage expert demonstrations, which represent superior car-following control experiences, to improve the DRL policy. Our method involves two sequential steps. In the first step, expert demonstrations are obtained during offline pretraining by utilizing prior traffic knowledge, including car-following trajectories from an empirical database and classic car-following models. In the second step, expert demonstrations are obtained during online training, where the agent interacts with the car-following environment. The EDRL agents are trained through supervised regression on the expert demonstrations using the behavioral cloning technique. Experimental results conducted in various traffic oscillation scenarios demonstrate that our proposed method significantly enhances training stability, learning speed, and rewards compared to baseline algorithms.

Neighboring human-driven vehicles (HDVs) frequently perform uncertain cut-in maneuvers, posing a threat to the safety and efficiency of autonomous vehicles (AVs), particularly in traffic oscillation scenarios characterized by AVs experiencing speed disturbances. In this paper, we propose an AV car-following strategy based on bounded rationality-aware reinforcement learning (BRARL) to handle cut-in maneuvers. The approach can handle scenarios involving simultaneous cut-in preclusion and cut-in yielding. Considering the limited rationality of human drivers during lane change, this strategy incorporates a bounded rationality-based game process to restrict discretionary cut-ins while safeguarding AV’s interests, including efficiency, safety, and comfort. The well-designed RL framework captures the exhibited randomness of both the cut-in and preceding HDVs, enabling the AV to effectively handle cut-in maneuvers and reduce speed disturbances. Simulated experiments demonstrate the high generalization capability of our strategy in reducing preceding speed disturbances (e.g., achieving a minimum reduction of 34.5% in traffic disturbances compared to two baselines), and preventing discretionary cut-in maneuvers.

Connected and automated vehicle (CAV) platooning exhibits significant potential in enhancing traffic efficiency and sustainability. In unsteady traffic conditions, CAV platoons frequently require splitting and merging maneuvers to avoid obstacles. This study introduces a hybrid automaton architecture for collision avoidance motion planning during CAV platoon merging and splitting. A velocity obstacle algorithm based on potential fields is developed to detect collision risks and calculate collision-free velocity solutions. Two predictive control-based optimization models are developed for collision-avoidance path planning, catering to both single-cruising vehicles and vehicle platoons. A synergetic architecture based on hybrid automaton is developed to coordinate vehicle motions during platoon splitting and merging. Numerical experiments are performed to evaluate the performance of the proposed hybrid automaton architecture under various obstacle scenarios. The results demonstrate that the proposed algorithms effectively identify collision risks within CAV platoons and determine optimal vehicle velocities. The proposed architecture demonstrates excellent performance in adjusting vehicle maneuvers and adapting CAV platoon formations to changing driving environments.

Traffic signal control (TSC) has been one of the most useful ways for reducing urban road congestion. The challenge of TSC includes 1) real-time signal decision, 2) the complexity in traffic dynamics, and 3) the network-level coordination. Reinforcement learning (RL) methods can query policies by mapping the traffic state to the signal decision in real-time, however, are inadequate for different traffic flow environment. By observing real traffic information, online planning methods can compute the signal decisions in a responsive manner. Unfortunately, existing online planning methods either require high computation complexity or get stuck in local coordination. Against this background, we propose an explicit multiagent coordination (EMC)-based online planning methods that can satisfy adaptive, real-time and network-level TSC. By multiagent, we model each intersection as an autonomous agent, and the coordination efficiency is modeled by a cost function between neighbor intersections. By network-level coordination, each agent exchanges messages of cost function with its neighbors in a fully decentralized manner. By real-time, the message-passing procedure can interrupt at any time when the real time limit is reached and agents select the optimal signal decisions according to current message. Finally, we test our EMC method in both synthetic and real road network datasets. Experimental results are encouraging: compared to RL and conventional transportation baselines, our EMC method performs reasonably well in terms of adapting to real-time traffic dynamics, minimizing vehicle travel time and scalability to city-scale road networks.

A promising way to improve efficiency in highway on-ramp regions is to control connected and automated vehicles (CAVs) to pass the merging section sequentially. The primary objective of this paper is to incorporate the reinforcement learning (RL) technique into vehicle merging control to achieve global cooperation. A communication protocol among RL agents is integrated with the Soft Actor-Critic (CSAC) algorithm. The parallel SAC agents with a parameter-sharing structure cooperate to optimize the common reward function. It enables CAVs to know the actions of each other so that they proactively negotiate to adjust speeds. We designed a parsimonious state representation containing crucial merging information to speed up the RL training. The effects of the merging control strategy were investigated with the simulation model in scenarios with different CAV penetration rates and traffic flow compositions. Results showed that the CSAC-based merging strategy generated collision-free merging trajectories with a short travel time while guaranteeing traffic safety in various traffic conditions. Four baselines (including two rule-based and two RL-based merging strategies) were applied in the same scenarios for comparison. It was found that the proposed merging strategy took the safest merging behaviors (zero traffic conflicts) and reduced the most significant travel time (56.9%).

The implementation of autonomous driving is inseparable from developing intelligent driving decision-making models, which are facing high scene complexity, poor decision-making coupling, and the inability to guarantee decision-making safety. This paper starts with the priority and logic of lane change and car-following decision-making, considering driving efficiency, safety, and comfort, then constructs a double-layer decision-making model. This paper uses two deep reinforcement learning algorithms for the upper and lower layers to process large-scale mixed state space and ensure the composite action output of lane-changing decisions and car-following decisions. In the upper layer model, we use the D3QN algorithm to distinguish the potential value of the environment and the value of selecting lane-changing actions when making lane-changing decisions. Different from the traditional mechanisms that only use negative rewards, the lane changing benefit function and dangerous action shielding mechanism are used to eliminate collisions. DDPG algorithm is adopted in the lower layer model to process car-following decisions and output continuous vehicle speed control. Besides, coupled training is taken for the two algorithms to improve the coordination of the double-layer model. This paper selected mixed standard driving cycle conditions to build a highly complex training environment and used NGSIM data to reconstruct scenes to test our model. Simulations in SUMO are presented that the double-layer model can increase the driving speed of the original data by 23.99%, which has higher effectiveness than other models.

In recent years, unmanned aerial vehicle (UAV) has become an increasingly popular tool for traffic monitoring and data collection on highways due to its advantage of low cost, high resolution, good flexibility, and wide spatial coverage. Extracting high-resolution vehicle trajectory data from aerial videos taken by a UAV flying over target highway segment becomes a critical research task for traffic flow modeling and analysis. This study aims at proposing a novel methodological framework for automatic and accurate vehicle trajectory extraction from aerial videos. The method starts by developing an ensemble detector to detect vehicles in the target region. Then, the kernelized correlation filter is applied to track vehicles fast and accurately. After that, a mapping algorithm is proposed to transform vehicle positions from the Cartesian coordinates in image to the Frenet coordinates to extract raw vehicle trajectories along the roadway curves. The data denoising is then performed using a wavelet transform to eliminate the biased vehicle trajectory positions. Our method is tested on two aerial videos taken on different urban expressway segments in both peak and non-peak hours on weekdays. The extracted vehicle trajectories are compared with manual calibrated data to testify the framework performance. The experimental results show that the proposed method successfully extracts vehicle trajectories with a high accuracy: the measurement error of Mean Squared Deviation is 2.301 m, the Root-mean-square deviation is 0.175 m, and the Pearson correlation coefficient is 0.999. The video and trajectory data in this study are publicly accessible for serving as benchmark at https://seutraffic.com.

The study aims to evaluate the performance of the transfer learning algorithm to enhance the transferability of a deep reinforcement learning-based variable speed limits (VSL) control. The Double Deep Q Network (DDQN)-based VSL control strategy is proposed for reducing total time spent (TTS) on freeways. A real merging bottleneck is developed in the simulation and considered for the VSL control as the source scenario. Three types of target scenarios are considered, including the overspeed scenarios, adverse weather scenarios, and diverse capacity drop scenarios. A stable testing demand and a fluctuating testing demand are adopted to evaluate the effects of VSL control. The results show that by updating the neural networks, the transfer learning in the DDQN-based VSL control agent successfully transfers knowledge learned in the source scenario to other target scenarios. With the transfer learning, the entire training process is shortened by 32.3% to 69.8%, while keeping a similar maximum reward level, as compared to the VSL control with full learning from scratch. With the transferred DDQN-based VSL strategy, the TTS is reduced by 26.02% to 67.37% with the stable testing demand and 21.31% to 69.98% with the fluctuating testing demand in various scenarios, respectively. The results also show that when the task similarity between the source scenario and target scenario is relatively low, the transfer learning could lead to local optimum and may not achieve the global optimal control effects.

Variable speed limits (VSL) is a major Intelligent Transportation System (ITS) technology for controlling freeway mainline traffic, which have been increasingly used to improve traffic safety and operations efficiency of freeway traffic management. The primary objective of this study is to evaluate the safety impacts of the VSL system implemented on Interstate 5 in Seattle, United States since 2010. A Full Bayesian (FB) before-after analysis was conducted based on 9,787 crashes that occurred in a 72-month study period. The analysis was conducted for all crashes, crash severity levels, crash types, and crash causes. The FB before-after results implied that the total crash count was reduced by 32.23% with a standard deviation of 3.58% after the VSL system was applied on the freeway. The count of crashes with no injury decreased more than crashes with severe injury and possible injury. The effect of rear-end crash reduction was the most beneficial among all crash types, while the effect on sideswipe crash reduction was the least. The study also compared the traffic speed features in the before and after periods to fully evaluate the impacts of the VSL system on traffic operations. The result indicated that, with VSL control, the difference in speed was reduced by the VSL system. The results of this study are particularly valuable for policy and control strategy development, and cost-benefit evaluation associated with VSL system implementations.

The vehicle platoon will be the most dominant driving mode on future roads. To the best of our knowledge, few reinforcement learning (RL) algorithms have been applied in vehicle platoon control, which has large-scale action and state spaces. Some RL-based methods were applied to solve single-agent problems. If we need to tackle multiagent problems, we will use multiagent RL algorithms since the parameters space grows exponentially with the increasing number of agents involved. Previous multiagent RL algorithms generally may provide redundant information to agents, indicating a large amount of useless or unrelated information, which may cause to be difficult for convergence training and pattern extractions from shared information. Also, random actions usually contribute to crashes, especially at the beginning of training. In this study, a communication proximal policy optimization (CommPPO) algorithm was proposed to tackle the above issues. In specific, the CommPPO model adopts a parameter-sharing structure to allow the dynamic variation of agent numbers, which can well handle various platoon dynamics, including splitting and merging. The communication protocol of the CommPPO consists of two parts. In the state part, the widely used predecessor-leader follower typology in the platoon is adopted to transmit global and local state information to agents. In the reward part, a new reward communication channel is proposed to solve the spurious reward and "lazy agent" problems in some existing multiagent RLs. Moreover, a curriculum learning approach is adopted to reduce crashes and speed up training. To validate the proposed strategy for platoon control, two existing multiagent RLs and a traditional platoon control strategy were applied in the same scenarios for comparison. Results showed that the CommPPO algorithm gained more rewards and achieved the largest fuel consumption reduction (11.6%).

Recently, the availability of unmanned aerial vehicle (UAV) opens up new opportunities for smart transportation applications, such as automatic traffic data collection. In such a trend, detecting vehicles and extracting traffic parameters from UAV video in a fast and accurate manner is becoming crucial in many prospective applications. However, from the methodological perspective, several limitations have to be addressed before the actual implementation of UAV. This paper proposes a new and complete analysis framework for traffic flow parameter estimation from UAV video. This framework addresses the well-concerned issues on UAV's irregular ego-motion, low estimation accuracy in dense traffic situation, and high computational complexity by designing and integrating four stages. In the first two stages an ensemble classifier (Haar cascade + convolutional neural network) is developed for vehicle detection, and in the last two stages a robust traffic flow parameter estimation method is developed based on optical flow and traffic flow theory. The proposed ensemble classifier is demonstrated to outperform the state-of-the-art vehicle detectors that designed for UAV-based vehicle detection. Traffic flow parameter estimations in both free flow and congested traffic conditions are evaluated, and the results turn out to be very encouraging. The dataset with 20,000 image samples used in this study is publicly accessible for benchmarking at http://www.uwstarlab.org/research.html.

Unmanned aerial vehicles (UAVs) are gaining popularity in traffic monitoring due to their low cost, high flexibility, and wide view range. Traffic flow parameters such as speed, density, and volume extracted from UAV-based traffic videos are critical for traffic state estimation and traffic control and have recently received much attention from researchers. However, different from stationary surveillance videos, the camera platforms move with UAVs, and the background motion in aerial videos makes it very challenging to process for data extraction. To address this problem, a novel framework for real-time traffic flow parameter estimation from aerial videos is proposed. The proposed system identifies the directions of traffic streams and extracts traffic flow parameters of each traffic stream separately. Our method incorporates four steps that make use of the Kanade-Lucas-Tomasi (KLT) tracker, k-means clustering, connected graphs, and traffic flow theory. The KLT tracker and k-means clustering are used for interest-point-based motion analysis; then, four constraints are proposed to further determine the connectivity of interest points belonging to one traffic stream cluster. Finally, the average speed of a traffic stream as well as density and volume can be estimated using outputs from previous steps and reference markings. Our method was tested on five videos taken in very different scenarios. The experimental results show that in our case studies, the proposed method achieves about 96% and 87% accuracy in estimating average traffic stream speed and vehicle count, respectively. The method also achieves a fast processing speed that enables real-time traffic information estimation.

The primary objective of this paper was to incorporate the reinforcement learning technique in variable speed limit (VSL) control strategies to reduce system travel time at freeway bottlenecks. A Q-learning (QL)-based VSL control strategy was proposed. The controller included two components: a QL-based offline agent and an online VSL controller. The VSL controller was trained to learn the optimal speed limits for various traffic states to achieve a long-term goal of system optimization. The control effects of the VSL were evaluated using a modified cell transmission model for a freeway recurrent bottleneck. A new parameter was introduced in the cell transmission model to account for the overspeed of drivers in unsaturated traffic conditions. Two scenarios that considered both stable and fluctuating traffic demands were evaluated. The effects of the proposed strategy were compared with those of the feedback-based VSL strategy. The results showed that the proposed QL-based VSL strategy outperformed the feedback-based VSL strategy. More specifically, the proposed VSL control strategy reduced the system travel time by 49.34% in the stable demand scenario and 21.84% in the fluctuating demand scenario.

This paper proposes a two-layer decision framework to model taxi drivers' customer-search behaviors within urban areas. The first layer models taxi drivers' pickup location choice decisions, and a Huff model is used to describe the attractiveness of pickup locations. Then, a path size logit (PSL) model is used in the second layer to analyze route choice behaviors considering information such as path size, path distance, travel time, and intersection delay. Global Positioning System data are collected from more than 36 000 taxis in Beijing, China, at the interval of 30 s during six months. The Xidan district with a large shopping center is selected to validate the proposed model. Path travel time is estimated based on probe taxi vehicles on the network. The validation results show that the proposed Huff model achieved high accuracy to estimate drivers' pickup location choices. The PSL outperforms traditional multinomial logit in modeling drivers' route choice behaviors. The findings of this paper can help understand taxi drivers' customer searching decisions and provide strategies to improve the system services.

The primary objective of this paper was to develop a control strategy of variable speed limits (VSLs) to reduce rear-end collision risks near freeway recurrent bottlenecks. The risks of rear-end collisions were estimated using a crash risk prediction model that is specifically developed for rear-end collisions in freeway bottleneck areas. The effects of the VSL control strategy were evaluated using a cell transmission model. Several control factors were tested, including the start-up threshold of the collision likelihood, the target speed limit, the speed change rate, and the speed difference between adjacent links. A genetic algorithm was used to optimize critical control factors. For the high demand scenario, the proposed control strategy used 25% of the maximum collision likelihood for the start-up threshold, 35 mi/h for the target speed limit, 10 mi/h per 30 s for the speed change rate, and 10 mi/h for the speed difference between different links. For the moderate demand scenario, the strategy used 20% of the maximum collision likelihood for the start-up threshold, 40 mi/h for the target speed limit, 15 mi/h per 30 s for the speed change rate, and 10 mi/h for the speed difference between different links. The results of comparative analyses suggested that the proposed control strategy outperformed other strategies in reducing the rear-end collision risks near freeway recurrent bottlenecks. With the proposed control strategy, the VSL control reduced the rear-end crash potential by 69.84% for the high demand scenario and by 81.81% for the moderate demand scenario.