Commercial platoon cruise control is an effective method to improve transportation efficiency, but existing research is mostly based on homogeneous platoon control with one single vehicle following optimization objective, while using a simple architecture to cope with communication time delay, which is not universally applicable in practical scenarios. Therefore, based on heterogeneous electric commercial vehicle fleets, in this paper a distributed model predictive control strategy is proposed to achieve multiobjective control that takes into account of requirements of vehicle following, economy, and comfort. Delay buffers and compensators are designed for delay prediction models, effectively solving the problems of excessive tracking distance error caused by nonideal communication conditions. Matlab/Simulink simulation shows that the proposed control algorithm can achieve multiobjective optimization control of heterogeneous commercial vehicle fleets. Compared with traditional model predictive control (MPC), it significantly reduces the tracking distance error, energy consumption, and jerk, effectively improving performance of the platoon in terms of tracking, economy and comfort and significantly reducing the adverse effect of time delay.

Traditional 3D object detection methods in Cartesian coordinate systems often overlook the symmetry and continuity of the target from different perspectives to some extent during camera image encoding due to the fixed wedgeshaped imaging geometry of invehicle cameras. To address this, in this paper, PolarDet, an innovative endtoend 3D object detection method in polar coordinates based on position and semantic information weighting is proposed. This method generates BEV (Bird's Eye View) position and semantic information in polar coordinates through polar coordinate queries and predefined polar grid, which then interacts with the BEV information from the previous frame to incorporate temporal information. When outputting the final detection results, PolarDet performs a weighted sum of position and semantic information to enhance information utilization efficiency, allowing the network to achieve higher detection accuracy. Extensive experiments on the challenging BEV object detection nuScenes dataset show that the optimal model of PolarDet achieves a mAP (mean average precision) of 0.469 and an NDS (nuScenes detection score) of 0.56, significantly outperforming Cartesian coordinatebased BEV detection methods.

In order to achieve customizable design and highfidelity intelligent driving simulation test perception data generation, an intelligent driving test scenario simulation architecture that integrates virtual and real perception data is established in this paper. By fusing simulated traffic subject perception data with real environment scene data, perception simulation data can be continuously generated with dangerous test scenarios as the target. On this basis, the RANSAC method is used to extract the position of obstacles in the real point cloud and determine the operating space constraints of simulated traffic subjects in the real environment scene at each moment. Then, in order to realize the interactive relationship between the behavior and position of the main vehicle and other traffic subjects in the test scenario, in the simulation software, simulation modeling and behavior design of the main vehicle and traffic subjects are conducted based on the real main vehicle sensor parameters and motion trajectories for output of continuous simulated traffic participant perception data. Finally, the mask replacement method and ray replacement strategy are used to perform virtual and real fusion on the image and point cloud data respectively, and the virtual and real fusion perception data of dangerous driving test scenes in different real environment scenarios are obtained. The simulation data is tested and verified. The results show that most scenarios in the real road collection data set have the ability to support simulation data injection. The injected simulated traffic subject behaviors can match the test scene requirements and have high authenticity. At the perceptual level, the injected simulated traffic subject and the real traffic subject have a similarity of 86.5% in the target detection algorithm confidence level. The proposed method can controllably inject simulated traffic subjects that meet test requirements into real environment scene data, and quickly and synchronously obtain virtualreal fusion images and point cloud data with high realism.

Given the high exposure and risk of rainfall as a trigger condition for visual perception systems, various rainfall simulation tests are the main research methods. However, the realism of rain simulation of different testing methods impacts the confidence in test conclusions. In this study indicators are selected to quantify the impact of rainfall on machine vision from the aspects of image quality and object detection. Using the numerical range and trend of index changes under real rainfall as a benchmark, the comparative study of the realism of different rainfall simulation methods in the dimension of machine vision is carried out. Additionally, in this study 1 950 images of no rain and various levels of real rainfall are collected to construct a dataset, so as to obtain statistical patterns of the impact of real rainfall on machine vision. Two simulated rainfall test sites, three simulation software, and one generative model are selected for rainfall simulation tests to compare and analyze the realism of different types of rainfall simulation methods horizontally. The results show that, in terms of image quality, simulation software and rainfall simulation equipment can better simulate the real rain in terms of DR value range and trend. Regarding target detection, simulation software and generative model are closer to real rainfall in terms of CC change values. Overall, in terms of realism, digital simulation of rainfall performs best, followed by physical rainfall simulation on site and generative model, providing a reference for testing the SOTIF of the visual perception system of intelligent and connected vehicles.

The vehicle stability region is an important aspect of research on vehicle stability analysis and control. For the problems of inaccurate description and difficult solution of stability region in existing research, a quadrilateral description and automatic solution method for vehicle stability region is proposed. A nonlinear twodegreeoffreedom vehicle model is established, and the ant colony algorithm is used to solve the equilibrium state of vehicle system. The Lyapunov indirect method is applied to determine the stability of the equilibrium state. Based on the phase plane of the sideslip anglesideslip angle velocity of mass center, several phase trajectory feature points and the phase plane stability region boundary point search method are established to solve the stability region boundary points. According to the different distributions of the vehicle stability region, two types of stability regions are proposed, and corresponding judgment methods, stability region quadrilateral description and its automatic solution methods are established. Based on the proposed method, the stability region of vehicle under common medium speed driving condition is solved. The results are compared with the parallel line method and diamond method, and the correctness of the quadrilateral description is validated by CarSim sine wave simulation results. The results show that the proposed quadrilateral description of the vehicle stability region can better describe the boundary of the stability region than the parallel line method and diamond method, and automatic solution reduces the workload of stability region solution.

To improve the handling stability of distributed drive electric vehicles (EVs) at high speeds on different road surfaces, in this paper an integrated control strategy for AFS/DYC based on hybrid model predictive control is proposed. Firstly, a piece affine tire model is constructed based on system identification methods. In conjunction with the vehicle dynamics model and the conversion relationship between propositional logic and linear inequalities, the vehicle system's mixed logical dynamic model is constructed. Then, an integrated control strategy for AFS/DYC based on hybrid model predictive control is designed. The strategy uses mixed integer quadratic programming to track target reference values for decisionmaking on additional yaw moment and additional steering angle, and constructs an optimized wheel driving torque distribution control strategy with the goal of minimizing tire load rate. Finally, a driverinloop handling stability test experiment is conducted on the CarSimSimulink cosimulation platform. The test results show that compared to the traditional model predictive control, the designed hybrid model predictive control strategy reduces the root mean square error of yaw rate and side slip angle by 31.61% and 19.51% respectively under highspeed double lane change conditions and the peak average torque amplitude of the four wheels is reduced by 24.27%.

With the rapid development of automotive active safety technology, the chassis electronic control unit of modern electric vehicles has seen explosive growth. In order to improve the realtime performance and accuracy of chassis active safety control, for the rapid growth of chassis electronic control units and the coupling conflict problems of low integration degree of control system and multiobjective cooptimization, in this paper firstly a chassis system integration control architecture based on multiagent is established, and a hierarchical control system integrating the front and rear wheels' active steering system and the differential braking control system is proposed. Secondly, based on this, the state equations of each agent and its contribution to the vehicle's center of mass model are established and combined with the model predictive control to consider the characteristics of constraints. The cost function containing global state tracking error and local control effort is designed considering both the actuator constraints and the ground friction ellipse constraints. Finally, each agent realizes its collaborative control through the interaction of dynamic information of its respective contribution. The results show that the vehicle stability control method based on multiagent model prediction proposed in this paper has obvious improvement in terms of traverse stability compared with independent control of each active safety unit under the driving conditions of high and low road attachment and large curvature curves, which has certain value for engineering application.

The torque distribution strategy plays a crucial role in improving the safety and energy efficiency of distributed drive electric vehicles. In order to reduce the energy consumption of electric vehicles with dualmotor drive on the front and rear axles, a multiobjective torque distribution method based on a hierarchical control architecture is proposed in this paper, that comprehensively considers vehicle safety, handling stability, and energy efficiency. The upper layer is the active safety layer, which uses nonlinear model predictive control (NMPC) to achieve vehicle safety and stability control. The lower layer is the torque distribution layer, which considers the torque control of the front and rear axle motors under noload loss of the motor. The simulation results show that compared with the average distribution method, the proposed multiobjective torque distribution method can improve the vehicle's stability while ensuring safe driving, with the total energy consumption reduced by 6.6% and 3.5% under the NEDC and WLTC driving cycles, respectively.

To solve the problem that a singlestage reduction hub drive system cannot meet the performance specifications of specialized vehicles, and that a twostage reduction would require the addition of extra control mechanisms, a novel coupled dual motor hub drive system scheme which is composed of two motors, a planetary gear mechanism, and a oneway clutch is proposed in this paper. The system is designed to operate in two coupling modes: torque coupling mode at low speed and speed coupling mode at medium and high speed with an autonomous mode switching capability as a functional requirement. Parameter matching is conducted with the objective of maximizing power utilization rate, and a control strategy for autonomous mode switching is developed. Under the same simulation initial condition, compared to the single motor two speed hub drive system, the coupled dual motor hub drive system exhibits an 81.25% reduction in maximum vehicle speed fluctuation and an 81.58% decrease in maximum acceleration during the mode switching process. An instantaneous optimal control strategy is established. Under the same operating conditions, the coupled dual motor hub drive system demonstrates a 21.42% reduction in energy consumption compared to the single motor twospeed hub drive system. Experimental tests are conducted using a prototype model to further validate the functionality and feasibility of the novel coupled dual motor hub drive system.

In this paper, an adaptive wheelbase preview robust H control method is proposed based on vibration based road roughness recognition to address the impact of unknown road surface input on the control effect of active suspension. By collecting the vibration acceleration response of the wheels through real vehicle experiments, the longitudinal road surface roughness information is identified based on the vibration based road surface roughness detection method of the front wheels. A speed adaptive wheelbase preview method is designed to obtain the delay relationship of the road surface excitation received by the front and rear wheels of the vehicle, providing real vehicle data for the wheelbase preview control of the rear wheel suspension. On this basis, a multiobjective speed adaptive wheelbase preview robust H. control method considering motion constraints is designed, and the optimal solution of parameters in linear matrix inequality (LMI) is achieved through multiobjective genetic algorithm (MOGA) to improve control accuracy. The experimental and simulation results show that the method proposed in this paper can accurately identify road roughness information and effectively improve suspension performance indicators and vehicle vibration frequency, effectively suppress vibration within the frequency range sensitive to motion sickness, and balance passenger driving experience while meeting driving smoothness requirements. Meanwhile, this method also provides a new approach for vertical vibration control of multi axle vehicles.