Simulation is an invaluable tool in the field of autonomous driving, especially in verifying the decision-making and planning algorithms. Real vehicle experiments can potentially cause safety accidents. In highly interactive driving scenarios, it is essential to verify whether autonomous driving vehicles intelligently interact with other traffic participants. In autonomous driving simulations, the environment vehicles’ motion model is crucial, and its intelligence directly affects the driving decisions of the test autonomous vehicles, which in turn determines the evaluation of the decision-making algorithm. The current research does not focus on modeling the motion of the environment vehicles, resulting in inaccurate evaluation results between the simulation and on-field experiment, where the real-world human drivers are more intelligent. This paper proposes a novel car-following model that considers the social preferences and predictive capabilities of real drivers based on the classical intelligent driver model. This model can adapt to social preferences after predicting the future trajectories of surrounding vehicles. When surrounding vehicles intend to change lanes, the social predictive intelligent driver model (SPIDM) can decelerate in advance, thereby enhancing both driving safety and comfort. In addition, real-world data are utilized to calibrate the SPIDM by extracting the car-following preferences of real drivers under the cut-in scenarios. Different categories of social preferences are obtained to generate diverse car-following behavior. Overall, SPIDM improves the intelligence level of environment vehicles, and creates more realistic traffic environment for the autonomous driving simulation.

This paper conducts a thorough exploration of vehicle-to-everything (V2X) communication in the realm of intelligent connected vehicles (ICVs). It initiates by tackling challenges across three pivotal phases of cooperative communication: pre-communication, during-communication, and post-communication. The discourse delves into a spectrum of concepts and strategies to surmount these challenges. Furthermore, it meticulously scrutinizes diverse communication scenarios and associated techniques, evaluating their significance and feasibility. Moreover, an in-depth analysis of various datasets is undertaken, considering their distinctive attributes and suitability for diverse communication tasks. The paper critically examines and debates the platforms and frameworks used in the experiments, providing valuable insights into their performance. Following a comprehensive review of existing methods and datasets, the paper identifies potential research directions and challenges that warrant further exploration in the realm of V2X communication for intelligent connected vehicles. This comprehensive examination contributes to a deeper understanding of the subject, paving the way for future advancements in this dynamic field.

This paper contributes to vehicle dynamics modeling by introducing a model architecture that addresses the state-space representations of nonlinear physical systems through data-driven techniques, thus bypassing the need for extensive parameterization. By effectively compressing time-series field test data, the model enables accurate long-term prediction of vehicle states and onboard sensor outputs, based on the sequential control inputs of driver assistance systems. This makes it suitable for both closed-loop simulation and offline analysis. Additionally, the model overcomes challenges such as sensor data unavailability and the effects of sensor noise in real-world scenarios. Utilizing a reduced-order approach within a neural state-space framework, the model comprises state and output networks configured as multi-layer perceptrons. The model is highly efficient in simulating vehicle dynamics during forward driving, its scalability is enhanced by a kinematics-based state limiter, which facilitates seamless transitions and improved adaptability across complex driving modes. The study also analyzes the impact of data down-sampling on model performance, which is crucial for the practical deployment of V2X. Validated through real-vehicle experiments under various driving conditions, the model demonstrates its capability to accurately replicate vehicle dynamics and its robustness in the face of sensor noise and down-sampled training data. This highlights the role of neural networks in refining both the credibility and practical utility of intelligent driving simulations. The source code used to train and evaluate the model is available at https://github.com/pansong/PyNSSM.

Intelligent vehicles require high positioning accuracy and stability. With the development of the internet of vehicles and intelligent transport system (ITS) and V2X (Vehicle to Everything) technology, vehicle positioning can now be achieved using a wider range of data sources. Traditional positioning methods are highly dependent on GNSS, while existing V2X positioning methods ignore the effect of latency. This paper proposes a V2X-based cooperative positioning method (CPM) through an enhanced KF and a transmission delay compensation to achieve accurate positioning. In CPM, the roadside perception of the vehicle is transmitted to the OBU through the V2X platform and is processed to reduce the effect of latency by historical IMU data to reconstruct the position. Then, an enhanced KF-based data fusion method is introduced, which utilizes a parallel KF and a Kalman gain adjustment strategy to fuse the compensated roadside information with the IMU data to achieve an optimal estimation of the vehicle position. Real vehicle experiments demonstrate that CPM improves the positioning accuracy and reduces the error caused by time delay. Compared to roadside lidar positioning, the RMSE of CPM is reduced by 34.7% and 59.1% on two test tracks, respectively. In addition, an ablation experiment verified the effectiveness of each module in CPM, and a comparison experiment demonstrated the superiority of the proposed fusion method.

The realization of personalized lane-changing (LC) for intelligent vehicles (IVs) is important for enhancing the social acknowledgment, user acceptance, adaptability, and trust of IVs. The LC style classification of human drivers represents a crucial foundation for achieving personalized LC. Therefore, this study constructs an LC style classification method based on driving behavioral primitives, which enables the classified LC styles to fully embody the implicit behavioral semantics and patterns of human drivers. First, a disentangled sticky hierarchical Dirichlet process hidden Markov model is proposed for the LC behavioral segment segmentation. The model can suppress frequent transitions of the hidden states, and vector autoregression is used to accurately describe the LC explicit behavioral parameters. Subsequently, the K-shape is employed to cluster all LC behavior segments to obtain interpretable and reasonable LC behavior primitives. Then, clustering features based on the LC behavioral primitives are constructed. Finally, LC styles are classified using density peak clustering, which does not require a manual specification of the number of clustering centers. Verification is performed on the Next Generation Simulation dataset, and the results indicate that this method can accurately and reasonably classify LC styles. The quantitative comparison with four state-of-the-art methods further demonstrates the advantages of the proposed method in LC style classification and confirms the effectiveness of introducing LC behavioral primitives.

A new type of transportation vehicle, the flying car, is attracting increasing attention in the automotive and aviation industries to meet people’s personalized transportation needs for urban air traffic and future travel. With its vertical take-off and landing capability, flying cars can expand its feasible routes into 3D space. The above process, however, requires sufficient path planning to obtain optimal 3D path. To solve the above issue, the inspiration was drawn from animals in the natural world to design a type of flying car that can travel in various urban environments such as land and low altitude by using different components like wheels and propellers. Incorporating the motion characteristics of flying cars in the future urban environment, segmenting the energy consumption and time models of various stages of flying cars is conducted. The introduction of temporal A* algorithm into the new field of flying cars for the first time, the priority planning algorithm for multiple flying car groups based on an improved A* algorithm utilizing safety intervals is proposed. The proposed strategy is validated on different sizes of urban environment maps. The results indicate that on a complex map with 452 nodes, the strategy effectively reduces distance by 4.5 m, decreases energy consumption by 85.8% and improves planning speed. Compared with the strategy based on multi-commodity network flow integer linear programming, the planning results are roughly the same, but the weighted cost of employing this strategy is decreased by 5.2%, and the path distance is reduced by 0.34 m.

Leading cruise control (LCC) in mixed traffic has received wide attention as it can strengthen the capability of connected and automated vehicles in reducing traffic instability and smoothing mixed traffic. However, existing predictive LCC cannot directly address head-to-tail string stability (HSS) since velocity fluctuation of human-driven vehicles behind cannot be handled by constraint design in MPC framework. To address this challenge, this paper proposes a control matching MPC approach for LCC in mixed traffic. A head-to-tail string stable feedback controller based on the inverse optimal velocity model is designed to guarantee HSS under bilateral topologies. Then, an MPC controller is proposed and the weighting matrices in the objective function are tuned to match the MPC controller with the head-to-tail string stable feedback controller. Straightforward analysis of HSS and physical/safety constraints satisfaction are neatly combined by the proposed control scheme. The feasibility and closed-loop stability of the MPC controller are analyzed. Finally, simulations verify the effectiveness of the proposed controller.

Accurately predicting the motion trajectories of surrounding vehicles is crucial for the safety of autonomous vehicles. Realizing the trajectory prediction of multi-target vehicles depends not only on the historical trajectories but also requires clarifying the dynamic spatial interactions between vehicles and the temporal relationships between trajectories of different time series. However, existing trajectory prediction methods do not adequately consider the coupling effects of spatial interactions and temporal relationships, resulting in insufficient accuracy for multi-target vehicles trajectory prediction in highly interactive scenarios. This paper proposes a planning-coupled multi-target vehicles trajectory prediction network (PCTP-Net) that contains encoding, feature fusion, and trajectory decoding modules for modeling coupled interactions based on time and space. Firstly, the encoding module employs a bidirectional long short-term memory (Bi-LSTM) network to encode historical and planning trajectories of different time series, which combines the planning information of the ego vehicle with the prediction process of multi-target vehicles to realize the coupled interaction modeling based on time and space. Secondly, the feature fusion module introduces a convolutional social pooling layer to analyze the impact of trajectories with different temporal features on the prediction and captures the dynamic spatial interactions between vehicles. Finally, the trajectory decoding module proposes a trajectory prediction decoder that incorporates driving behavior decisions to improve the trajectory prediction accuracy of multi-target vehicles under interaction. The experiments on the NGSIM dataset and different traffic scenarios show that the proposed method can achieve accurate trajectory prediction in traffic scenarios with dense and highly interactive vehicles.

Energy management strategies (EMSs) play a critical role in determining the performance of hybrid electric vehicles (HEVs). In the context of an off-road HEV equipped with two power sources—an engine-generator set and a hybrid energy storage system—one of the key challenges is optimizing energy distribution while balancing multiple objectives. This paper addresses this challenge by developing a multi-objective strategy based on a novel multi-agent reinforcement learning algorithm, specifically the Nash deep Q-network (Nash DQN). The proposed EMS seeks to improve fuel efficiency, maintain the battery state of charge, and extend battery lifespan while meeting the ultracapacitor SOC constraint. The Nash DQN algorithm, which integrates concepts from game theory and deep reinforcement learning, is employed to solve the multi-objective optimization problem. In this framework, the EGS and HESS are treated as individual agents, and their interactions are modeled as general-sum stochastic games. Through the application of Nash DQN’s theory, the two agents generate optimal actions that ensure Nash equilibrium, thus achieving a balanced trade-off between the various objectives. Finally, the performance of Nash DQN is evaluated through simulations, where it is compared to conventional DQN and dynamic programming approaches. The results demonstrate the superiority of Nash DQN in addressing the multi-objective optimization for HEV energy management.

A yaw moment control strategy for four wheel independent drive electrics vehicle is proposed in this paper. The control strategy is a hierarchical architecture which containing a yaw motion generation layer and a longitudinal force distribution layer. The yaw motion generation layer consists of feedforward control and feedback control. Unlike previous strategy, in this paper, yaw rate is considered as the only control variable which is feasible to be detected in practice. The feedforward control is used to enhance overshoot in transient condition while the feedback control is to change vehicle steady state steering characteristic and extend its lateral limit. The longitudinal force distribution layer is designed based on vehicle wheel load transfer model. It makes each tire fully utilized its lateral force limits. Based on these two layers, a control model is built accordingly. Both steady state and transient state experiments are conducted in Carsim simulation associated with the built Simulink control model. The simulation results show that proposed control strategy can enhance vehicle handling performance in steady state and transient state. Experiments were conducted on an electric vehicle which evaluated the accuracy of the established model and the effect of the proposed yaw moment control strategy.

During the shifting process of hybrid transmission, the motor plays an important role due to its fast response and accurate torque control. However, it also leads to an excessive reliance on the motor, while neglecting the collaborative control of the motor and clutch. Therefore, there is still room for further improvement in the shifting quality of the hybrid system. In addition, the changing motor torque can easily cause a certain shifting impact at the moment of the clutch state transition, especially for the transition from sliding state to engaged state. In the meanwhile, due to the presence of multiple power sources, the influence of acceleration on the shifting process cannot be ignored. For these problems, this paper proposes a power-on upshifting control strategy for a two-speed series-parallel hybrid transmission, and the root cause of the shifting impact when the clutch slip is eliminated is analyzed. On this basis, a motor torque control method is proposed that is easily to be implemented in engineering, that is, when the target slip is reduced to the threshold ?ω1, set the motor proportional-integral (PI) controller output to zero to ensure that the motor output torque can be changed to zero when the clutch slip is eliminated so that the shifting impact here can be suppressed. Then the factors that affect the value setting of Δω1 are analyzed based on the dynamics model and verified by simulation and hardware-in-loop (HIL) test based on the control variable method. In addition, the influence of acceleration on the shifting process is analyzed, and corresponding control strategies are proposed to improve the shifting quality. Finally, the Worldwide Harmonized Light Vehicles Test Cycle (WLTC) test is conducted on the vehicle model to compare the shifting control effects under different working conditions, which can verify the proposed effectiveness of the control strategy.

Accurate trajectory tracking control with guaranteed vehicle dynamics stability is fundamental for automated vehicles. This paper proposes an enabling trajectory tracking scheme with improved dynamics stability for four-wheel-independent-drive electric vehicles. First, a modified phase plane method is proposed for assessing vehicle dynamics stability by constructing a 3-D phase trajectory. Then a lateral trajectory tracking controller is developed using a model predictive control algorithm with a variable stiffness tire model. Considering the effect of the additional yaw moment on the trajectory tracking error and vehicle yaw stability, a dual-weight cooperative sliding mode control method is established for yaw angle tracking compensation and yaw stability control through target switching. Finally, the desired additional yaw moment and total longitudinal tire force are achieved through wheel torque allocation while accounting for the slip ratios. The performance of the proposed method is evaluated through comprehensive Hardware-in-the-Loop tests under double-lane change maneuvers.