Automotive Innovation最新文献

Estimating depth from images captured by camera sensors is crucial for the advancement of autonomous driving technologies and has gained significant attention in recent years. However, most previous methods rely on stacked pooling or stride convolution to extract high-level features, which can limit network performance and lead to information redundancy. This paper proposes an improved bidirectional feature pyramid module (BiFPN) and a channel attention module (Seblock: squeeze and excitation) to address these issues in existing methods based on monocular camera sensor. The Seblock redistributes channel feature weights to enhance useful information, while the improved BiFPN facilitates efficient fusion of multi-scale features. The proposed method is in an end-to-end solution without any additional post-processing, resulting in efficient depth estimation. Experiment results show that the proposed method is competitive with state-of-the-art algorithms and preserves fine-grained texture of scene depth.

The state-of-the-art deep neural networks are vulnerable to the attacks of adversarial examples with small-magnitude perturbations. In the field of deep-learning-based automated driving, such adversarial attack threats testify to the weakness of AI models. This limitation can lead to severe issues regarding the safety of the intended functionality (SOTIF) in automated driving. From the perspective of causality, the adversarial attacks can be regarded as confounding effects with spurious correlations established by the non-causal features. However, few previous research works are devoted to building the relationship between adversarial examples, causality, and SOTIF. This paper proposes a robust physical adversarial perturbation generation method that aims at the salient image regions of the targeted attack class with the guidance of class activation mapping (CAM). With the utilization of CAM, the maximization of the confounding effects can be achieved through the intermediate variable of the front-door criterion between images and targeted attack labels. In the simulation experiment, the proposed method achieved a 94.6% targeted attack success rate (ASR) on the released dataset when the speed-speed-limit-60 km/h (speed-limit-60) signs could be attacked as speed-speed-limit-80 km/h (speed-limit-80) signs. In the real physical experiment, the targeted ASR is 75% and the untargeted ASR is 100%. Besides the state-of-the-art attack result, a detailed experiment is implemented to evaluate the performance of the proposed method under low resolutions, diverse optimizers, and multifarious defense methods. The code and data are released at the repository: https://github.com/yebin999/rp2-with-cam.

Cooperative adaptive cruise control (CACC) is an important technology for improving road utilization and energy efficiency in the automotive industry. In CACC systems, connected vehicles can receive information from adjacent ones through communication networks. However, the networks are vulnerable to cyber-attacks, so the states of vehicles cannot be received promptly and accurately. This paper studies the security resilience control for a CACC system subject to denial of service (DoS) attack. The core of the proposed resilient control strategy is to estimate the delay caused by DoS attack and then compensate for it in the controller. Specifically, a CACC system is modeled by considering the impacts of DoS attack on the transmitted data. Then, a high-gain observer is presented to estimate the vehicle states including the time delay. The convergence of the observer is proved in a theorem based on the Lyapunov stability theory, and the high-gain-velocity observer is modified so that the estimation error of the velocity can converge to zero in a finite time. A resilient controller is designed by proposing a time delay compensation algorithm to mitigate the impacts of DoS attack. The effectiveness of the estimation and control methods is illustrated by a ten-vehicle simulation system operating at the FTP75 driving cycle conditions. And the relative estimation errors are less than 6%.

Electric vehicles are developing prosperously in recent years. Lithium-ion batteries have become the dominant energy storage device in electric vehicle application because of its advantages such as high power density and long cycle life. To ensure safe and efficient battery operations and to enable timely battery system maintenance, accurate and reliable detection and diagnosis of battery faults are necessitated. In this paper, the state-of-the-art battery fault diagnosis methods are comprehensively reviewed. First, the degradation and fault mechanisms are analyzed and common abnormal behaviors are summarized. Then, the fault diagnosis methods are categorized into the statistical analysis-, model-, signal processing-, and data-driven methods. Their distinctive characteristics and applications are summarized and compared. Finally, the challenges facing the existing fault diagnosis methods are discussed and the future research directions are pointed out.

Convective heat transfer plays an important role in the development of a high-performance battery cell. Electric vehicles carry a large amount of the battery cells to reach a longer range of endurance mileage. Thermal diffusion around the battery cells can be considered as obstacles to improve the convective heat transfer coefficient. In this paper, a novel agitator taking advantage of strong vortices is designed to disrupt the thermal boundary layer around the battery cells, thereby improving the fluid mixing for enhanced convective heat transfer. A fluid–structure interaction algorithm is developed to simulate the convective heat transfer rate at various flapping motion. Under the comparison with clean channel, the vortex-induced vibration by the agitated beam can increase the average Nusselt number by 119.59%. This research can be applied to optimize the thermal-structure design inside the electric vehicle battery.

An objective evaluation scheme for automotive technical and comprehensive performance could provide critical and instructive insights for academic research, engineering practice, and commercial marketing of vehicles. In this paper, the technical performance index (A = S/left( {T_{1} cdot T_{2} } right){ }) (({text{m}}/({text{s}}^{2} cdot ;{text{L}}))) and comprehensive performance index (F = M cdot S/left( {T_{1} cdot T_{2} } right),) (({text{kN}}cdot;{text{L}}^{ - 1}), where (M) is the vehicle mass) are formulated by incorporating the vehicle 0–100 ( {text{km}}cdot;{text{h}}^{ - 1}) acceleration duration ({ }T_{1}), 100–0 ( {text{km}}cdot;{text{h}}^{ - 1}) braking duration ({ }T_{2}), and fuel economy (S) (mileage per liter fuel at constant speed) to assess the vehicle’s longitudinal dynamic performance. (A) and (F) offer a clear physical implication of a vehicle’s acceleration capability and traction efficiency acquired per unit of fuel consumption, respectively. These indexes are used for wide case studies of popular market sedans and SUVs of joint ventures (JVs) and domestic brands in China over the last 17 years. The findings prove that this approach could be effectively and reliably utilized for the objective evaluation and analysis of the technical and comprehensive performance of automotive models.

Automated valet parking (AVP) has attracted the attention of industry and academia in recent years. However, there are still many challenges to be solved, including shortest path search, optimal time efficiency, and applicability of algorithm in complex scenarios. In this paper, a hierarchical AVP path planner is proposed, which divides a complete AVP path planning into the guided layer and the planning layer from the perspective of global decision-making. The guided layer is mainly used to divide a complex AVP path planning into several simple path plannings, which makes the hybrid A* algorithm more applicable in a complex parking environment. The planning layer mainly adopts different optimization methods for driving and parking path planning. The proposed method is verified by a large number of simulations which include the verification of the optimal parking position, the performance of the planner for perpendicular parking, and the scalability of the planner for parallel parking and inclined parking. The simulation results reveal that the efficiency of the algorithm is increased by more than 20 times, and the average path length is also shortened by more than 20%. Furthermore, the planner overcomes the problem that the hybrid A* algorithm is not applicable in complex parking scenarios.

The fuel cell hybrid powertrain is a potential power supply system for fuel cell vehicles. The underlying problem is that the fuel cell vehicles encounter exhaustive hydrogen consumption. To effectively manage hydrogen consumption, the aim is to propose fuel cell city bus power and control system. The underlying idea is to determine the target power of fuel cell through simulation study on fuel cell and battery energy management strategy and road test verifications. A half-power prediction energy management strategy is implemented to predict the target power of the fuel cell in the current time step based on the demand power of the vehicle and the state of charge (SOC) of the battery in the previous time steps. This offers better understanding of the correlation between fuel cell power and vehicle drive cycle for enabling effective power supply management. The research results show that the half-power prediction energy management strategy effectively reduces the hydrogen consumption of the vehicle by 7.1% and the number of battery cycle by 6.0%, compared to the stepped management strategy of battery SOC. When applied to a 12-m fuel cell city bus—F12, specially designed and manufactured for the Winter Olympic Games in 2022—the fuel economy of 3.7 kg/100 km is achieved in urban road conditions. This study lays a foundation for providing the powertrain configuration and energy management strategy of fuel cell city bus.

The driver's behavior plays a crucial role in transportation safety. It is widely acknowledged that driver vigilance is a major contributor to traffic accidents. However, the quantitative impact of driver vigilance on driving risk has yet to be fully explored. This study aims to investigate the relationship between driver vigilance and driving risk, using data recorded from 28 drivers who maintain a speed of 80 km/h on a monotonous highway for 2 hours. The k-means and linear fitting methods are used to analyze the driving risk distribution under different driver vigilance states. Additionally, this study proposes a research framework for analyzing driving risk and develops three classification models (KNN, SVM, and DNN) to recognize the driving risk status. The results show that the frequency of low-risk incidents is negatively correlated with the driver's vigilance level, whereas the frequency of moderate-risk and high-risk incidents is positively correlated with the driver's vigilance level. The DNN model performs the best, achieving an accuracy of 0.972, recall of 0.972, precision of 0.973, and f1-score of 0.972, compared to KNN and SVM. This research could serve as a valuable reference for the design of warning systems and intelligent vehicles.