Automotive Innovation最新文献

This paper proposes a novel density-based real-time segmentation algorithm, to extract ground point cloud in real time from point cloud data collected by roadside LiDAR. The algorithm solves the problems such as the large amount of original point cloud data collected by LiDAR, which leads to heavy computational burden in ground point search. First, point cloud data is filtered by straight-through filtering method and rasterized to improve the real-time performance of the algorithm. Then, the density of the point cloud in horizontal plane is calculated, and the threshold of the density is selected to extract the low-density regional point cloud according to the density statistical histogram and 95% loci. Finally, the low-density regional point cloud is used as the initial ground seeds for iterative optimization of ground parameters, and the ground point cloud is extracted by the fitted ground model to realize road point cloud extraction. The experimental results on 1055 frames of continuous data collected on real scenes show that the average time consumption of the proposed method is 0.11 s, and the average segmentation precision is 92.48%. This shows that the density-based road segmentation algorithm can reduce the time of point cloud traversal in the process of ground parameter fitting and improve the real-time performance of the algorithm while maintaining the accuracy of ground extraction.

Driver distraction has been deemed a major cause of traffic accidents. However, drivers’ brain response activities to different distraction types have not been well investigated. The purpose of this study is to investigate the response of electroencephalography (EEG) activities to different distraction tasks. In the conducted simulation tests, three secondary tasks (i.e., a clock task, a 2-back task, and a navigation task) are designed to induce different types of driver distractions. Twenty-four participants are recruited for the designed tests, and differences in drivers’ brain response activities concerning distraction types are investigated. The results show that the differences in comprehensive distraction are more significant than that in single cognitive distraction. Friedman test and post hoc two-tailed Nemenyi test are conducted to further identify the differences in band activities among brain regions. The results show that the theta energy in the frontal lobe is significantly higher than that in other brain regions in distracted driving, whereas the alpha energy in the temporal lobe significantly decreases compared to other brain regions. These results provide theoretical references for the development of distraction detection systems based on EEG signals.

Clustering is an unsupervised learning technology, and it groups information (observations or datasets) according to similarity measures. Developing clustering algorithms is a hot topic in recent years, and this area develops rapidly with the increasing complexity of data and the volume of datasets. In this paper, the concept of clustering is introduced, and the clustering technologies are analyzed from traditional and modern perspectives. First, this paper summarizes the principles, advantages, and disadvantages of 20 traditional clustering algorithms and 4 modern algorithms. Then, the core elements of clustering are presented, such as similarity measures and evaluation index. Considering that data processing is often applied in vehicle engineering, finally, some specific applications of clustering algorithms in vehicles are listed and the future development of clustering in the era of big data is highlighted. The purpose of this review is to make a comprehensive survey that helps readers learn various clustering algorithms and choose the appropriate methods to use, especially in vehicles.

The take-over control (ToC) of human–machine interaction is a hotspot. From automatic driving to manual driving, some factors affecting driver response time have not been considered in existing models, and little attention has been paid to its effects on mixed traffic flow. This study establishes a ToC model of response based on adaptive control of thought-rational cognitive architecture (CAR-ToC) to investigate the effects of driver response time on traffic flow. A quantification method of driver’s situation cognition uncertainty is also proposed. This method can directly describe the cognitive effect of drivers with different cognitive characteristics on vehicle cluster situations. The results show that when driver response time in ToC is 4.2 s, the traffic state is the best. The greater the response time is, the more obvious the stop-and-go waves exhibit. Besides, crashes happen when manual vehicles hit other types of vehicles in ToC. Effects of driver response time on traffic are illustrated and verified from various aspects. Experiments are designed to verify that road efficiency and safety are increased by using a dynamic take-over strategy. Further, internal causes of effects are revealed and suggestions are discussed for the safety and efficiency of autonomous vehicles.

The advanced and personalised experience that modern cars offer makes them more and more data-hungry. For example, the cabin preferences of the possible drivers must be recorded and associated to some identity, while such data could be exploited to deduce sensitive information about the driver’s health. Therefore, drivers’ privacy must be taken seriously, requiring a dedicated risk assessment framework, as presented in this paper through a double assessment combining the asset-oriented ISO approach with the threat-oriented STRIDE approach. The framework is tailored to the level of specific car brand and demonstrated on the ten top-selling brands as well as, due to its innovative character, Tesla. The two approaches yield different, but complementary findings, demonstrating the additional insights gained through their parallel adoption.

Increasing efficiency and reducing emissions are fundamental approaches to achieving peak carbon emissions and carbon neutrality for the transportation and power industries. The Argon power cycle (APC) is a novel concept for high efficiency and zero emissions. However, APC faces the challenges of severe knock and low power density at high efficiency. To elevate efficiency and power density simultaneously of APC, the Miller cycle is applied and combined with APC. The calculation method is based on a modification of the previous thermodynamic method. The mixture of hydrogen and oxygen is controlled in the stoichiometric ratio. The results indicate that to obtain a thermal conversion efficiency of 70%, in the Otto cycle, the compression ratio and the AR (argon molar ratio in the argon-oxygen mixture) could be 9 and 95%, respectively. In comparison, for the Miller cycle, these two parameters only need to be 7 and 91%. A lower compression ratio can reduce the negative effect of knock, and a reduced AR increases the power density by 66% with the same efficiency. The improvement effect is significant when the expansion-compression ratio is 1.5. Meanwhile, increasing the expansion-compression ratio is more effective in the argon-oxygen mixture than in the nitrogen–oxygen mixture. For the next-generation Argon/Miller power cycle engine, the feasible design to achieve the indicated thermal efficiency of 58.6% should be a compression ratio of 11, an expansion-compression ratio of 1.5, and an AR of 91%.

This paper presents an energy management optimization system based on an adaptive functional state model of battery aging for internal combustion engine vehicles (ICEVs). First, the functional characteristics of batteries in ICEVs are investigated. Then, an adaptive functional state model is proposed to represent battery aging throughout the entire battery service life. A battery protection scheme is developed, including over-discharge and graded over-current protection to improve battery safety. A model-based energy management strategy is synthesized to comprehensively optimize fuel economy, battery life preservation, and vehicle performance. The performance of the proposed scheme was examined under comprehensive test scenarios based on field and bench tests. The results show that the proposed energy management algorithm can effectively improve fuel economy.

A novel real-time predictive control strategy is proposed for path following (PF) and vehicle stability of autonomous electric vehicles under extreme drive conditions. The investigated vehicle configuration is a distributed drive electric vehicle, which allows to independently control the torques of each in-wheel motor (IWM) for superior stability, but bringing control complexities. The control-oriented model is established by the Magic Formula tire function and the single-track vehicle model. For PF and direct yaw moment control, the nonlinear model predictive control (NMPC) strategy is developed to minimize PF tracking error and stabilize vehicle, outputting front tires’ lateral force and external yaw moment. To mitigate the calculation burdens, the continuation/general minimal residual algorithm is proposed for real-time optimization in NMPC. The relaxation function method is adopted to handle the inequality constraints. To prevent vehicle instability and improve steering capacity, the lateral velocity differential of the vehicle is considered in phase plane analysis, and the novel stable bounds of lateral forces are developed and online applied in the proposed NMPC controller. Additionally, the Lyapunov-based constraint is proposed to guarantee the closed-loop stability for the PF issue, and sufficient conditions regarding recursive feasibility and closed-loop stability are provided analytically. The target lateral force is transformed as front steering angle command by the inversive tire model, and the external yaw moment and total traction torque are distributed as the torque commands of IWMs by optimization. The validations prove the effectiveness of the proposed strategy in improved steering capacity, desirable PF effects, vehicle stabilization, and real-time applicability.

Precise state and parameter estimations are essential for identification, analysis and control of vehicle engineering problems, especially under significant model and measurement uncertainties. The widely used filtering/estimation algorithms, such as Kalman series like Kalman filter, extended Kalman filter, unscented Kalman filter, and particle filter, generally aim to approach the true state/parameter distribution via iteratively updating the filter gain at each time step. However, the optimality of these filters would be deteriorated by unrealistic initial condition or significant model error. Alternatively, this paper proposes to approximate the optimal filter gain by considering the effect factors within infinite time horizon, on the basis of estimation-control duality. The proposed approximate optimal filter (AOF) problem is designed and subsequently solved by actor-critic reinforcement learning (RL) method. The AOF design transforms the traditional optimal filtering problem with the minimum expected mean square error into an optimal control problem with the minimum accumulated estimation error, in which the estimation error is used as the surrogate system state and the infinite-horizon filter gain is the control input. The estimation-control duality is proved to hold when certain conditions about initial vehicle state distributions and policy structure are maintained. In order to evaluate of the effectiveness of AOF, a vehicle state estimation problem is then demonstrated and compared with the steady-state Kalman filter. The results showed that the obtained filter policy via RL with different discount factors can converge to theoretical optimal gain with an error within 5%, and the average estimation errors of vehicle slip angle and yaw rate are less than 1.5 × 10^–4.

The full-lifespan management concept provides a new pathway to seeking solutions from macro-application scenarios to micro-mechanism levels. This paper presents a cyber hierarchy multiscale optimal control method for multiple intelligent hybrid vehicles to fully release the potentials of vehicle components while guaranteeing driving safety and stability. It can be generally divided into the cyber intelligent driving system on the cyber-end and the intelligent vehicle system on the vehicle-end. On the cyber-end, the state information of the surrounding vehicles is transmitted via the Vehicle-to-Everything structure and further processed in the cloud platform to generate future driving behaviors based on a car-following theory. On the vehicle-end, an optimized control sequence for vehicle components at micro-levels is derived by incorporating a physics-informed neural network model for battery health prediction. The results show that global optimization needs high coupling between the macro- and micro-physical processes. By introducing the genetic algorithm for time smoothing, the improved driving strategy is capable of macro- and micro-coupling, and thus improves the controllable performance in time series. Moreover, this method spans the complexity of space, time, and chemistry, enhances the interpretation performance of machine learning, and slows down the battery aging in the process of multiscale optimization.