In the context of real-world driving environments，due to the perturbation of perception data and the unpredictable behavior of other traffic participants，rational decision-making in highly interactive and intricate driving scenarios considering the impact of uncertainty factors is one of the main concerns that decision-making and planning systems for autonomous vehicles must address. A behavioral decision-making method for autonomous vehicles navigating in uncertain environments is proposed in this paper. To mitigate the impact of uncertainty，the behavioral decision-making process is transformed into a partially observable Markov decision process （POMDP）. Furthermore，to tackle the computational complexity of the POMDP model，the complex network theory is applied for the first time for dynamically modeling the microscopic driving environment surrounding the autonomous vehicle，which allows for the effective characterization of interaction relationship between vehicle nodes and the scientific selection of significant vehicle nodes，guiding the autonomous vehicle's decision-making process，enabling precise identification of critical vehicle nodes，and pruning the decision space. The effectiveness of the proposed method is verified in a simulation environment，and the experimental results show that the proposed method has higher computational efficiency，superior performance，and enhanced flexibility in comparison to existing state-of-the-art behavioral decision-making methods.

To address the impact of sparsity and disorder of point clouds on target detection accuracy，a two-stage multimodal fusion network VPC-VoxelNet based on virtual point clouds is proposed in this paper. Firstly，virtual point clouds are constructed using image detection target information to increase the density of point clouds，thus improving the performance of target features. Secondly，the dimensionality of point cloud features is increased to distinguish real and virtual point clouds，and a voxel with confidence encoding is used to enhance the correlation of point clouds. Finally，the scale factor of the virtual point clouds is adopted to design the loss function to increase the supervised training of image detection and improve the training efficiency of the two-stage network，and avoid the cumulative model error problem of the two-stage end-to-end network model. The target detection network，VPC-VoxelNet，is tested on the KITTI dataset，and the detection accuracy is better than that of the classical 3-dimensional point cloud detection network and certain multi-sensor information fusion networks，with a vehicle detection accuracy of 86.9%.

In shared road space，human driving interaction behavior has the social characteristics of considering the impact on surrounding vehicles. Lacking the understanding of such social characteristics，autonomous vehicles often struggle to estimate the potential impact of their behavior on surrounding vehicles，thus falling into over conservativeness of decision-making dilemma. A game-theory-based social driving interaction model is constructed by introducing in the behavioral characteristics of drivers considering the impact on surrounding vehicles to capture the action dependencies among road users. With this model，a generalized measurement，utility term of interaction activeness （UTIA），is proposed to quantify the potential impact of the host vehicle's anticipated behavior on its interactants. By introducing the UTIA into the planning objective，the interaction activeness of motion planning algorithm can be directionally adjusted. The results of highway exit experiments show that without compromising safety，enhancing interaction activeness can improve the success rate of the exit task within a given distance by 3.9% and 5.2% for optimization-based and sampling-based motion planning algorithm，respectively.

In order to improve the traffic efficiency and fuel utilization efficiency of intelligent connected vehicles （ICVs） under urban traffic networks，a multilane spatiotemporal trajectory optimization method is proposed in this paper. Firstly，the state and constraints of the ICVs are defined based on the multi-lane spatiotemporal position relationship and the compound optimization model of spatiotemporal trajectory is constructed by considering the traffic efficiency and fuel economy，which is solved by the Pontryagin Maximum algorithm. Furthermore，the rules of cooperative lane change are designed to obtain the optimal lane change strategy by Q-learning algorithm. Finally，the SUMO/Python co-simulation tests show that the method can effectively improve the traffic efficiency under different vehicle saturation levels，split allocation，and minimum traffic speed conditions，with great improvement of fuel efficiency.

Carbon reduction in commercial vehicles has become a key bottleneck in reducing carbon emission in China's road transportation. New energy commercial vehicles are seen as an important way to reduce carbon emission in heavy commercial vehicles，but the market penetration rate of new energy commercial vehicles is much lower than that of other vehicle sectors. However，at present，the development of new energy zero-emission commercial vehicles still faces significant bottlenecks such as complex application scenarios，diversified technological paths，and high cost. This study constructs a Discrete Choice-based Market Evolution of Green Truck Model （DC-MEGT），a multi-dimensional Logit discrete choice model based on factors such as the total cost of ownership （TCO） and ease of use of new energy vehicles. TCO is calculated using a bottom-up approach，and the usage convenience is quantified and monetized by supplementary energy time cost. A comprehensive utility function is constructed to predict and analyze the market penetration rate evolution of different power types，such as pure electric vehicles，fuel cell vehicles，and zero-emission fuels from the present to 2060. The study analyzes the heavy-duty long-haul towing scenario as an example and finds that the main technology paths in 2060 include fuel cell vehicles，pure electric vehicles，natural gas vehicles，and diesel vehicles，accounting for 48%，28%，12%，and 10%，respectively. If the uncertainty of different factors such as policy promotion，technological progress，and business models is taken into account，the market share of pure electric vehicles and fuel cell vehicles in 2060 may fluctuate by 17% ~ 19%.

The weak cold-start capability of fuel cells with graphite plates for vehicles is an important bottleneck that affects the large-scale promotion of fuel cell vehicles in the cold regions of northern China. Starvation self-heating is a common cold-start strategy whose basic principle is to increase overpotential by reducing the supply rate of reactants，and generate a large amount of heat inside the cell in a short period of time to achieve rapid heating. This approach is simple，but it requires a high degree of consistency in the initial water content of the stack monomers and is prone to single-chip reverse polarity and excess hydrogen concentration emission，which can affect the safety and durability of the fuel cell. To solve the above problems，the research group has developed a multi-channel AC impedance measurement device，proposed an optimized purging strategy for single cell impedance consistency，and established a constant voltage and variable air flow control method for cold-start of fuel cells，to achieve multi-objective and multi-parameter coupled coordinated control that provides high heat production，high safety，and high dynamics for voltage，current，and inlet/outlet air flow in the low-temperature start transient process. The bench test results show that the maximum impedance deviation of fuel cells is decreased from 0.7 to less than 0.2 mΩ，and the fuel cell engine system can achieve a fast start at -40 ℃ within 124 s，with good repeatability. The relevant technology is applied in the fuel cell demonstration at the 2022 Winter Olympics，with its effectiveness verified.

In order to optimize the cold start process of PEMFC，it is essential to provide sufficient feedback data. Common impedance spectroscopy and equivalent circuits cannot provide sufficient and real-time feedback due to long acquisition period. Therefore，the cold start impedance model is developed in COMSOL，and the change of impedance spectroscopy is analyzed in combination with experiments. The characteristic frequencies of 1kHz，50Hz and 1Hz are proposed in the high，medium and low frequency ranges respectively to characterize the cold start process of the fuel cell. The results show the above characteristic frequencies vary significantly in the pre-，mid- and post-cold start phases，with the change in impedance at characteristic frequencies of 1 kHz，50Hz and 1Hz of 0.38，0.31 and 1.47 respectively. It improves the real-time performance of data acquisition while retaining feature information compared to obtaining the full impedance spectroscopy and fitting the equivalent circuit. Therefore，the impedance at the characteristic frequency points can be used to characterize the cold start process，which provides real-time monitoring for the internal state of the cold start.

In view of the complex and changeable vehicle environment of electric vehicles，which affects the measurement accuracy of current sensors，and the worse situation will lead to the failure of one-phase or multi-phase current sensors in the motor drive system，therefore，a non-current-sensor control algorithm based on extended Kalman filter is proposed in this paper. The stator current of the motor is reconstructed by using the stator voltage，rotor position and speed information of the permanent magnet synchronous motor，and the feed forward compensation is designed to improve the dynamic performance of the system regarding the system delay caused by the non-current-sensor algorithm. The acceleration and deceleration and robustness tests of the proposed algorithm are carried out. The effectiveness of the proposed method is verified by the simulation and experimental results.

Cell to body （CTB） is a key technology to improve the endurance mileage of electric vehicles. The VRB/OW-GFRP hybrid structure formed by the variable-thickness rolled blanks （VRB） structure and the orthotropic woven GFRP （OW-GFRP） through the bonding process is an innovative structure，which can reduce the weight of the CTB battery pack，thus improve the endurance mileage of electric vehicles. Taking an electric vehicle as the research object，a CTB battery body integrated structure is designed to realize the integration of the upper cover of the battery pack and the floor of car body. Furthermore，the cover assembly of the uniform thickness （UT） CTB battery pack are replaced by the VRB structure，UT/OW-GFRP and VRB/OW-GFRP hybrid structure. The lightweight design of the upper cover assembly of the three types of CTB battery pack are carried out based on the multi-stage optimization method. The results show that the weight of VRB structure is reduced by 6.4% compared with that of UT structure when the stiffness performance of the CTB battery pack is met. The lightweight level of the upper cover assembly of the CTB battery pack based on the VRB/OW-GFRP hybrid structure is about three times that of the metal structure，with the weight of the VRB/OW-GFRP reduced further by 4.2% compared with that of the UT/OW-GFRP hybrid battery pack upper cover assembly. Thus，the VRB/OW-GFRP hybrid structure is the inevitable trend of the development of automotive lightweight technology in the future，showing a great application prospect in CTB battery pack cover assembly.

To address the problem of high communication overhead and low security of key distribution in the near-field communication of VANETs，an efficient quantum group key distribution scheme based on quantum random numbers is proposed in this paper. In this scheme，Firstly，the anonymous credentials of the vehicle are jointly generated by quantum random numbers at the vehicle side and the cloud side，and zero-knowledge proof is used to achieve mutual identity recognition between the vehicle and the road side，which protects the privacy of the vehicle. Then，a two-stage group key achieves the update of the group key，i.e. two key parameters at the roadside and the cloud side. The group key reduces the signaling overhead by half and greatly shortens the group key issuance time while ensuring forward and backward security. Finally，the security of the scheme is demonstrated by security analysis and performance analysis.