In the Social Internet of Vehicles (SIoV), federated learning is able to significantly protect the private data of the vehicle’s client, while reducing the transmission load between entities. Nevertheless, data can still be stolen by an adversary who analyzes the parameters uploaded by the client to steal it. In this paper, to effectively prevent data leakage and reduce the propagation delay of data, we design a federated learning collaborative authentication protocol for shared data. The parameters of the vehicle client model are encrypted by the protocol in the federated learning. The vehicle and other entities of the protocol realize efficient anonymous mutual authentication and key agreement. The security of the proposed protocol is proved in the stochastic predictive machine model. The simulation results on the SUMO and OMNeT++ platforms show that the authentication delay is the lowest compared to other protocols and the packet loss rate is reduced to 4.68%. Moreover, the overfitting of the globally aggregated model is effectively resolved.

Propelled by advancements in deep learning technology, autonomous driving has undergone rapid development in recent years. Object sensing, as the core technology of autonomous vehicles, is undergoing a transformation from traditional single-modal sensing to multi-modal sensing, aiming to perceive environmental information in a robust, accurate, and real-time manner to adapt to increasingly complex application scenes. This paper aims to offer a comprehensive guide for radar-camera fusion, focusing on the object detection task. Firstly, detailed explanations are provided regarding the characteristics, principles, data processing methods, and representations of radar and camera data. Secondly, we analyze the evaluation metrics of object detection via a group-independent form and summarize existing radar-camera fusion datasets. Thirdly, we comprehensively review and analyze key issues in radar-camera fusion across five hierarchies (why, what, where, when, and how to fuse). Fourthly, we discuss and anticipate potential challenges and emerging research directions from the view of the robustness, accuracy, and real-time performance for object detection.

In recent years, with the increasing severity of energy and environmental issues, countries have vigorously developed the new energy automotive industry. To reduce the difficulty of driver operation and increase endurance mileage, this article proposes a regenerative braking control strategy for a single-pedal pure electric commercial vehicle. Firstly, the single-pedal control system’s hierarchical approach was designed to contain the driver’s intention analysis and torque calculation layers. After identifying the driver’s intention, a logic threshold method was used to determine the braking pattern. Then, a fuzzy theory was used, with road gradient, braking strength, and speed as input parameters, and the ratio coefficient of braking force as the output parameter. A hybrid regenerative braking strategy was formulated based on the ideal distribution curve. Finally, the proposed strategy was verified through simulation and a constant-speed car-following experiment. The constant-speed car-following experiment results show that the maximum optimization rate of energy consumption provided by the proposed single-pedal regenerative braking control strategy is 5.81%, and the average optimization rate is 4.33%. This strategy can effectively reduce energy consumption and improve the economic performance of single-pedal pure electric commercial vehicles.

Federated learning has been extensively utilized in distributed learning scenarios for the Internet of Vehicles. However, two key challenges exist: 1) Gradients are frequently transmitted between vehicles during FL training, which reduces the communication timeliness between the traffic participants. 2) The fixed gradient compression method cannot sufficiently adapt to the dynamic IoV network topology. Therefore, we design an adaptive vehicle clustering method constructed according to multiple attributes, such as computational resources, communication distance, and latency. Accordingly, we propose a dynamic gradient compression strategy that filters similar local training models between vehicles and uses the Wasserstein distance to compute a sparsity threshold. This threshold acts as a dynamic compression factor that compresses gradient model parameters, reducing redundant parameter transmission. Furthermore, we conduct experiments using two datasets to evaluate the proposed strategy’s effectiveness. The compression ratio improved by 132-and 178-fold compared to the baselines, and the aggregated accuracy increased by an average of 10.13%. Additionally, experiments incorporating communication noise revealed that the aggregation model of the signal noise ratio is -29 dB.

Narrow parking spaces pose difficulties in path planning and spot turning caused by sudden changes and discontinuities in path curvature. To address these problems, this paper investigates the performance of three path-planning algorithms and proposes a path-optimization algorithm. First, a narrow parking space is defined based on single-step parking using the arc-line combination parking algorithm. Second, to compare the arc-line combination algorithm, Hybrid A* algorithm, and particle swarm optimization parking algorithm with respect to different narrow parking spaces, a multi-objective evaluation function is proposed, including three evaluation indicators, namely, the path length, the number of positive and negative conversions of vehicle speed, and the smoothness of the path. Their performance is compared using a simulation conducted in MATLAB. With the same starting point and different parking space widths, the three algorithms are simulated to generate different planned paths. Then, the evaluation indices are obtained to compare the performance of the algorithms based on the multi-objective function, the values of which indicate the fitness of the algorithm in a narrow parking environment. The results show that the Hybrid A* algorithm is better than the others for narrow parking spaces. Third, to smooth the planning path, a path optimization algorithm based on the cubic B-spline curve and gradient descent is proposed. Finally, the results of a simulation conducted on the proposed algorithm and the Hybrid A* algorithm are provided: the average minimum curvature of the path was reduced by 0.005 m⁻¹, and the path meets the requirements of the minimum turning radius constraint of the analyzed vehicle. The results show that the proposed algorithm can effectively eliminate the curvature mutation point and constrain the path curvature to meet the requirements of a smooth path.