In recent years, reinforcement learning (RL) methods have shown powerful learning capabilities in single-vehicle autonomous driving. However, few studies have focused on multi-vehicle cooperative driving based on RL, particularly in the dynamically changing traffic environments of highway ramp merge zones. In this paper, a multi-agent deep reinforcement learning (MARL) framework for multi-vehicle cooperative decision-making is proposed based on actor–critic, which categorizes vehicles into two groups according to their origins in the merging area. At the same time, the complexity of the network is reduced and the training process of the model is accelerated by utilizing mechanisms such as partial parameter sharing and experience playback. Additionally, a combination of global and individual rewards is adopted to promote cooperation in connected autonomous vehicles (CAVs) and balance individual and group interests. The training performance of the model is compared under three traffic densities, and our method is also compared with state-of-the-art benchmark methods. The simulation results show that the proposed MARL framework can have stronger policy learning capability and stability under various traffic flow conditions. Moreover, it can also effectively improve the speed of vehicles in the merging zone and reduce traffic conflicts.

The booming growth of the internet of things has brought about widespread deployment of devices and massive amounts of sensing data to be processed. Federated learning (FL)-empowered mobile edge computing, known for pushing artificial intelligence to the network edge while preserving data privacy in learning cooperation, is a promising way to unleash the potential information of the data. However, FL’s multi-server collaborative operating architecture inevitably results in communication energy consumption between edge servers, which poses great challenges to servers with constrained energy budgets, especially wireless communication servers that rely on battery power. The device-to-device (D2D) communication mode developed for FL turns high-cost and long-distance server interactions into energy-efficient proximity delivery and multi-level aggregations, effectively alleviating the server energy constraints. A few studies have been devoted to D2D-enabled FL management, but most of them have neglected to investigate server computing power for FL operation, and they have all ignored the impact of dataset characteristics on model training, thus failing to fully exploit the data processing capabilities of energy-constrained edge servers. To fill this gap, in this paper we propose a D2D-assisted FL mechanism for energy-constrained edge computing, which jointly incorporates computing power allocation and dataset correlation into FL scheduling. In view of the variable impacts of computational power on model accuracy at different training stages, we design a partite graph-based FL scheme with adaptive D2D pairing and aperiodic variable local iterations of heterogeneous edge servers. Moreover, we leverage graph learning to exploit the performance gain of the dataset correlation between the edge servers in the model aggregation process, and we propose a graph-and-deep reinforcement learning-based D2D server pairing algorithm, which effectively reduces FL model error. The numerical results demonstrate that our designed FL schemes have great advantages in improving FL training accuracy under edge servers’ energy constraints.

Traffic breakdown refers to the first-order transition from free flow to synchronized flow. It is characterized by a rapid decrease in speed, suddenly increasing density, and abruptly plummeting capacity in most of the relevant observations. To understand its cause and model its empirical observations, a multi-regime car-following model is proposed, which classifies the car-following state into four regimes, i.e., Free driving, High-speed following, Low-speed following, and Emergency. Simulation results demonstrate that traffic breakdown and spontaneous jam formation can be reproduced simultaneously by the new model. Experimental verification has shown that the new model can successfully simulate the observed concave growth pattern of traffic oscillations.

In existing traffic simulation software or studies related to traffic flow at signalized intersections, the treatment of yellow lights is often simplified or overlooked. However, driving behavior during the yellow phase is characterized by significant uncertainty, which can lead to discrepancies between simulation results and real-world conditions. To address this issue, this paper develops a driving behavior model based on risk perception and fuzzy decision-making and integrates it into a simulation framework to replicate continuous driving behaviors at isolated signalized intersections. The performance of intersections under varying yellow light durations is analyzed, yielding some key findings. For instance, when vehicles strictly adhere to the designed speed, increasing the yellow light duration from 3 s to 5 s results in higher traffic volumes under high traffic density. Furthermore, real-time traffic speed fluctuations stabilize, and the occurrence of unsafe driving behaviors decreases. The concept of risk perception is employed to explain the underlying mechanisms behind these phenomena. This paper provides both a theoretical foundation and a simulation framework for more detailed representations of driving behaviors and for explaining the fundamental principles governing intersection performance.

With the rapid development of intelligent driving technology, in order to meet the high-precision positioning requirements of intelligent vehicles, this paper considers the problem that high-speed moving vehicle mounted cameras can easily produce blurry or blurred images, leading to difficulties in extracting QR code information and affecting recognition accuracy. Therefore, a QR road sign detection and recognition technology based on Yolov5 and GAN is proposed. Firstly, the object detection method based on deep learning YOLOv5 is used to detect and locate QR code images; Then, GAN is used to enhance the target image, generate clear QR codes, and finally decode the QR image to obtain information and obtain the QR code information. By identifying five indicator parameters including target quantity, detection rate, effective recognition distance, recognition rate, and detection time, and comparing them with open-source tools such as Pyzbar and WeChat scanning engine. Experiments have shown that during high-speed driving, the detection and recognition rates of this method are both over 95%. It has similar detection and recognition rates to the other two mature methods, but the number of targets detected by this method in the same road environment is about twice that of other methods, and the effective recognition distance is longer. This indicates that this method helps high-speed vehicles obtain more roadside QR code information and can obtain higher accuracy vehicle positioning information. Therefore, this method can effectively meet the high-precision requirements of vehicles under high-speed operating conditions while meeting the detection and recognition rates. It is also suitable for vehicle positioning needs in scenarios such as tunnels where GPS signals are interfered with or lost.