Search

Image corruptions due to noise, blur, contrast change, etc., could lead to a significant performance decline of Deep Neural Networks (DNN), which poses a potential threat to DNN-based autonomous vehicles. Previous works attempted to explain corruption from a Fourier perspective. By comparing the absolute Fourier spectrum difference between corrupted images and clean images in the RGB color space, they regard the noise from some corruptions (Gaussian noise, defocus blur, etc.) as concentrating on the high-frequency components while others (contrast, fog, etc.) concentrate on the low-frequency components. In this work, we present a new perspective that unifies corruptions as noise from high frequency and thus propose an image augmentation algorithm to achieve a more robust performance against common corruptions. First, we notice the 1/fα statistical rule of the natural image's spectrum and the channels-wise differential sensitivity on the YCbCr color space of the Human Visual System. Thus we present a new perspective of the relative Fourier spectrum in the YCbCr color space which unifies the noise from all 15 common corruptions as noise concentrating on the high-frequency components. From the new perspective, the SOTA (state-of-the-art) image augmentation algorithm shows insufficient coverage of frequency change compared with common corruptions and an insufficient improvement in robustness against perturbations on the frequency domain. Second, based on the unified understanding of image corruptions and limitations of the SOTA image augmentation algorithm, we present a frequency-based image augmentation that adds noise to the high-frequency components in the YCbCr color space with the amplitude of noise proportional to the spectrum amplitude of the image. Compared with previous works, the proposed method achieves SOTA performance against corruptions on common datasets (CIFAR-10 and tiny-ImageNet) and real-world driving tests.

Low-frequency vibration caused by road roughness while driving is transmitted to the human body through tires, suspension, and seats. Prolonged exposure of the human body to the vibratory environment will have an impact on ride comfort or even health issues. In order to investigate the vibration response of various segments of occupants while driving, a 15-DOF multi-body dynamic model depicting the shanks with feet, thighs, pelvis, torso with arms, and the head of occupants is established in the two-dimensional sagittal plane, which considers the contact between the occupant and the cushion, backrest headrest, and the vehicle floor simultaneously. The biodynamic parameters are obtained by fitting the published vibration experimental data based on an optimization algorithm. The previously proposed half-car model is incorporated into the human model to construct an integrated vehicle-human model for further ride comfort analysis. The bump road and the random road are respectively adopted as the displacement excitation of the model to simulate the vibration response of each segment of occupants. The contact stiffness and damping between the human body and seats are modified and a sensitivity analysis is performed to explore the influence of various seat parameters on ride comfort. By changing the speed of the vehicle, we investigate the discrepancies in ride comfort on a given road at different speeds. This study reveals the vibration characteristics of seated occupants, vehicles, and the integrated vehicle-human system, based on which a method for evaluating the ride comfort of occupants is established.

High-speed vehicles in low illumination environments severely blur the images used in object detectors, which poses a potential threat to object detector-based advanced driver assistance systems (ADAS) and autonomous driving systems. Augmenting the training images for object detectors is an efficient way to mitigate the threat from motion blur. However, little attention has been paid to the motion of the vehicle and the position of objects in the traffic scene, which limits the consistence between the resulting augmented images and traffic scenes. In this paper, we present a vehicle kinematics-based image augmentation algorithm by modeling and analyzing the traffic scenes to generate more realistic augmented images and achieve higher robustness improvement on object detectors against motion blur. Firstly, we propose a traffic scene model considering vehicle motion and the relationship between the vehicle and the object in the traffic scene. Simulations based on typical ADAS test scenes show that the high vehicle speed and near object position is the key factor in generating motion blur. Second, we propose the vehicle-motion-based image augmentation algorithm. The proposed method applies the motion blur on the clear object based on the vehicle's speed and the relative position of the object. Subjective evaluation and multiple objective evaluation indexes including structural similarity index measure (SSIM), perceptual hash, normalized mutual information, and cosine similarity demonstrates that the proposed image augmentation can produce images more consistent with the traffic scenes. Thirdly, we apply the proposed method to the training of object detectors. Experiments on the KITTI dataset as well as real-world driving tests show that the proposed image augmentation achieves a higher robustness improvement than existing image augmentation algorithms on multiple object detectors including CenterNet, YOLOv3, and Faster R-CNN.

Brake creep groan is getting more and more attention in vehicle brake design and development. However, many complex motions of brake creep groan have not been comprehensively, which are critical for the understanding and prevention of the problem. In this paper, brake creep groan dynamics are studied and analysed to research the mechanisms of creep groan. Vehicle road experiments on a slope road are performed to record creep groans using two tri-axial accelerometers, which are install at the caliper and suspension strut. History curves of oil pressure and caliper accelerations, time–frequency spectrogram of the vibrations, and largest Lyapunov exponent are calculated using Hilbert–Huang transform and empirical mode decomposition methods. This study reveals that the sources of the brake creep groan include unstable sliding in addition to transient and steady-state motions of stick–slip between disc and pad. It is found that one of the important sources of the creep groan is the unstable sliding, which has impulsive and discontinuous nature and has effects of hammering leading to strong excitation and wideband vibrations. It is also found that when creep groan occurs, multiple modes of the brake system and subsystems contribute to the complex vibrations. The existences and the properties of the stable, unstable, and chaotic vibrations associated with the creep groan are characterized, which depend on the system natures, excitations, and tribological conditions.

In-vehicle active sound quality control (ASQC) system is an advanced application of active noise control (ANC) technology, which always suppresses the interior sound pressure or improves the psychoacoustic features. However, the connection between the sound quality control state and the vehicle's working state is normally ignored, and the passenger's sound perception typically expects a linear relationship between the vehicle speed and sound loudness. This paper presents a dual sampling-rate active noise equalization algorithm. The low sampling-rate signal is used to improve system computing efficiency, the high sampling-rate signal is used to improve the ASQC effect, and their conversion communication uses a hold (low to high) and a sampler (high to low). The data relating to the ASQC system was collected, and some sound quality evaluations were conducted for determining two appropriate sampling-rate values. Combining a genetic algorithm and the corresponding ASQC simulation system, the algorithmic optimal convergence coefficients and gain coefficients were further determined at different engine speeds. The verification results of ASQC using this proposed algorithm show that the loudness of interior noise is effectively suppressed; the nonlinear index is reduced to 1.37, improving by 37% relative to the original noise and by 20% compared to the ANC system.

Friction and contact characteristics play important roles in brake squeal generation, while their heterogeneous characteristics are often neglected. In this paper, a flexible pin-on-disc system model is established. Squeal bench test is carried out to verify the correctness of the model. Random heterogeneous friction coefficients and contact stiffnesses are built and assigned to the model. Evaluation indexes of friction and contact heterogeneities are established. The unstable region and the squeal uncertainty are analyzed. Then, the influence of the heterogeneous friction and contact characteristics on generation of brake squeal is explored based on statistical method and uncertainty concept in the unstable region and the mix region respectively. The results show that when only friction coefficient is heterogeneous, the radius of the center position of friction coefficient is the key influencing factor of brake squeal; when only contact stiffness is heterogeneous, the radius of the center position of contact stiffness is the key influencing factor of brake squeal; when friction coefficient and contact stiffness are both heterogeneous, the mean value of the product of friction coefficient and contact stiffness and the radius of the center position of friction coefficient play critical roles on friction squeal. Besides, squeal tends to occur when the mean value of the product of heterogeneous friction coefficient and contact stiffness and the radius of the center position of friction coefficient are large.

To determine the interior noise reduction mechanism of car body cavity filler block, the noise reduction researches were conducted experimentally. By comparing the BIW modal characteristic and transfer functions before and after removing cavity filler block, the influence of cavity filler block on vehicle low frequency vibration was investigated. To investigate the effect of cavity filler block on engine noise and road-tire noise, a full vehicle test using drum tester was carried out. Another full vehicle test in the wind tunnel was conducted to determine how it influences wind noise. Based on the above test results, the noise reduction mechanisms were analyzed. On the one hand, it helps to increase the damping of BIW, thus suppress the vibration and reduce low frequency noise. On the other hand, it cuts off the propagation of the noise outside the vehicle which may invades the passenger compartment, thereby reducing the high frequency noise.

In this paper, we investigate vehicle–bridge interactions and propose a new method for assessing bridge conditions based on nonlinear vibration analysis. A vehicle model is presented, and the dynamic load on the bridge is quantified based on bridge road roughness. The vibration of a river bridge under a moving vehicle is recorded and analyzed. The Fourier spectrum of the bridge response due to vehicle loading exhibits complicated cluster bands in addition to multiple peaks, some of which are correlated with the bridge natural modes characterized by finite element method analysis and free decay testing, and some of which are correlated with the vehicle dynamic load spectrum. To characterize the unidentified narrow band in the bridge response spectrum, the Lyapunov exponent of the response is estimated, which is found to be a positive value indicating the nonlinear properties of the bridge’s large response under vehicle loading. This finding correlates with the bridge inspection, which shows that some supporting roller bearings do not properly contact the superstructure of the bridge. To elaborate this phenomenon, a model of beam with nonideal supporting condition is analyzed, and a narrow band spectrum is characterized using He’s variational-iteration method. This research offers insight to using the vehicle–bridge interaction to monitor the health status of a bridge by using nonlinear properties, as well as linear properties.

Based on the theory of random response analysis, a road noise prediction method is proposed. A finite element model of acoustic-structure coupling for the full-vehicle was established, and the road surface texture measured in the actual vehicle test was converted into a power spectrum, which is the excitation of FE model. Under the conditions of coast down from 80km/h to 20km/h on Belgian pavement, the prediction of road noise in the 20- 200Hz frequency band was achieved. The driver’s ear sound pressure obtained from the simulation is in good agreement with the test results, which verifies the effectiveness of the simulation model and prediction method. The sensitivity of the driver’s ear-side sound pressure level to the stiffness of the suspension mounts was further analysed. The stiffness of the bushings that had a large impact was determined. The Response Surface Methodology was used to optimize road noise. The results show that the average sound pressure level near the driver’s ear is reduced by 1.81 dBA. The simulation analysis and optimization methods introduced in this paper can provide a reference for the virtual development process of vehicle road noise.

This paper investigates the creep groan of a vehicle’s brake experimentally, analytically, and numerically. Experimentally, the effects of acceleration on caliper and strut, noise, brake pressure, and tension are measured. The results show that the measured signals and their relevant spectra broadly capture the complex vibrations of creep groan. This includes the simple stick-slip, severe stick-slip vibrations/resonances, multiple harmonics, half-order harmonics; stick-slip-induced impulsive vibrations, steady/unstable vibrations, and their transitions. Analytically, a new mathematical model is presented to capture the unique features of half-order harmonics and the connections to fundamental stick-slip/resonant frequency and multiple harmonics. The analytical solution and the experimental results show that the vibro-impact of the brake pad-disc system can be triggered by severe stick-slip vibrations and is associated with instable, impulsive stick-slip vibration with wideband. The induced stick-slip vibro-impact can evolve into a steady and strong state with half-order, stick-slip fundamental, and multiple-order components. This new mechanism is different from all previously proposed mechanisms of creep groan in that we also view some type of creep groan as a stick-slip vibration-induced vibro-impact phenomenon in addition to conventional stick-slip phenomena. The new mechanism comprehensively explains the complex experimental phenomena reported in the literature. Numerically, the salient features of phase diagrams of instable stick-slip and vibro-impact are examined by using a seven-degree-of-freedom brake system model, which shows that the phase diagrams of the dynamics of creep groan with and without vibro-impact are substantially different. The phase diagram of the dynamics with vibro-impact is closer to the experimental results. In contrast to existing mechanisms, the proposed new mechanism encompasses the instable stick-slip nature of creep groan and elaborates the inherent connections and transition of the spectrogram. The new knowledge can be used to attain critical improvements to brake noise and vibration analysis and design. By applying the proposed new model in addition to existing models, all experimental phenomena in creep groan are elaborated and quantified.

To solve the problem that the mainstream capacity allocation method of driving and braking systems for distributed drive electric vehicles adopts uniform distribution among four wheels, a novel capacity allocation method of driving system and braking system is proposed. Among them, the capacity allocation of driving system for power performance is to improve the maximum acceleration, the maximum gradability, and the 0–100 km/h acceleration time; the capacity allocation method of electro-hydraulic braking system for braking performance is to improve the braking distance and the average braking deceleration. The results of the theoretical analysis indicate that the proposed capacity allocation method effectively improves the power performance and braking performance of the vehicle.

Brake noise is the hotspot in the field of automotive NVH. Till now, four main frictional squeal mechanisms for the generation of brake noise are highlighted; they are mode coupling instability theory, negative friction-velocity slope instability theory, sprag-slip motion theory, and stick-slip motion theory. However, none of them can explain various frictional squeal phenomena fully. So, it is very important to make clear the relationship among various mechanisms. A nonlinear dynamical friction-induced vibration model with two-point contact and three-degree-of-freedoms is established to explore the mechanism of frictional squeal and the relationship between several main mechanisms. Based on this model, bifurcations of friction coefficient for mode coupling are derived under zero, negative and positive friction-velocity slope conditions. According to the bifurcation friction coefficient for mode coupling and the sign of friction-velocity slope, instability characteristics are summarized in four areas. For each area, real part and imaginary part of instable mode and stick-slip motion are investigated via numerical simulation. At the same time, divergence is calculated and the existence of limit cycle is judged based on Bendixson–Dulac criterion. Based on these results, the relationship among the four mechanisms is more clearly understood.