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. more...

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. more...

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. more...

Autonomous driving related technologies have become a hot topic in academia and industry. Planning control is one of the core technologies of autonomous driving, which is conducive to vehicles safe and efficient driving. This paper proposes a novel optimal speed control algorithm, which considers the power system's energy consumption, the speed limit on the road, and the safe distance of the vehicle in front. An optimal speed control model of “From battery to wheel” energy consumption is established by constructing a performance index function based on the best-fitting formula of motor power, motor speed and torque. Based on the optimal control principle, the fourth-order ordinary differential equation of the speed control model is established, based on the indirect adjoining approach, the speed control model under the restriction of the road speed limit and safe distance of the preceding vehicle is derived and the analytical expression is obtained. Under the restriction of the road speed limit and safe distance of the preceding vehicle, the model predictive control method is introduced, the urban working conditions in the NEDC cycle are selected for simulation, and the optimal control input is solved in real-time, compared with the intelligent driver driving strategy, this algorithm is proved to be superior in the aspect of vehicle speed optimization control. Finally, the influence mechanism of the road speed limit constraint on the vehicle speed control is analyzed. This paper realizes the design of the optimal speed control algorithm for energy consumption under various constraints. The results can be used to guide the energy-saving driving of vehicles, and provide theoretical guidance for ecological driving technology in the context of intelligent transportation. more...