Search Results

As a key component of in-vehicle intelligent voice technology, speech enhancement can extract clean speech signals contaminated by environmental noise to improve the perceptual quality and intelligibility of speech. It has extensive applications in the field of intelligent car cabins. Although some end-to-end speech enhancement methods based on time domain have been proposed, there is often limited consideration given to designing model architectures based on the characteristics of the speech signal. In this paper, we propose a new U-Net based speech enhancement framework that utilizes the temporal correlation of speech signals to reconstruct higher-quality and more intelligible clean speech.

Speech enhancement can extract clean speech from noise interference, enhancing its perceptual quality and intelligibility. This technology has significant applications in in-car intelligent voice interaction. However, the complex noise environment inside the vehicle, especially the human voice interference is very prominent, which brings great challenges to the vehicle speech interaction system. In this paper, we propose a speech enhancement method based on target speech features, which can better extract clean speech and improve the perceptual quality and intelligibility of enhanced speech in the environment of human noise interference. To this end, we propose a design method for the middle layer of the U-Net architecture based on Long Short-Term Memory (LSTM), which can automatically extract the target speech features that are highly distinguishable from the noise signal and human voice interference features in noisy speech, and realize the targeted extraction of clean speech.

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.

Driver behaviors provide abundant information and feedback for future Advanced Driver Assistance Systems (ADAS). Driver’s eye and head may present some typical movement patterns before executing driving maneuvers. It is possible to use driver’s head and eye movement information for predicting driver intention. Therefore, to determine the most important features related to driver intention has attracted widespread research interests. In this paper, a method to analyze the correlation between driver’s visual features and driver intention is proposed, aiming to determine the most representative features for driver intention prediction. Firstly, naturalistic driving experiment is conducted to collect driver’s videos during executing lane keeping and lane change maneuvers. Then, driver’s head and face visual features are extracted from those videos. By using boxplot and independent samples T-test, features which have significant correlation with driver intention are found.

It’s significant to analyze the Experimental performance of active control system for road noise. In this paper, a 2-channel active control system of vehicle road noise based on FXLMS algorithm is established. The complexity of Filtered-x Least Mean Square algorithm (FXLMS) is analyzed. The bench test and road test are carried out to test and analyze the performance of the control system. Firstly, the general mathematic model of the multi-channel active control system based on FXLMS algorithm is established. The computational complexity of the algorithm is analyzed. Secondly, a hardware-in-the-loop (HIL) test bench based on multi-channel FXLMS algorithm and a measurement system based on DASP are set up, to measure the noise reduction performance of active noise control system under various working conditions. Finally, the bench test and the road test are carried out and the results are analyzed.

As the key assembly of new energy vehicles, the noise and vibration, and harshness (NVH) performance of integrated electric drive system directly affects the driving quality of new energy vehicles. In this paper, the vibration noise characteristic test of 3in1 electric drive system is carried out in the semi-muffler chamber. In order to compare and analyze the difference between 2in1 and 3in1 electric drive system NVH performance, the power electronics unit (PEU) in the 3in1 system was removed and placed on the ground away from the platform, and vibration noise test was carried out. In order to analyze the difference of NVH performance between 2in1 status and 3in1 status, the PEU in the 3in1 system was removed and placed on the ground far away from the bench, and the NVH test was carried out. The microphone signal at 1m position and the vibration acceleration signal of the key structural surface of the system are measured experimentally.

The parameter setting has a great influence on the noise reduction performance of the road noise active control (RNC) system. This paper analyzes and optimizes the parameters of the RNC system. Firstly, the model of the RNC system is established based on the FxLMS algorithm. Based on this model, taking the maximum noise reduction as the evaluation index, the sensitivity analysis of convergence coefficient, filter order, and reference signal gain was carried out using the Sobol method with the data measured by a real vehicle on asphalt pavement at 40km/h. The results show that there is no significant interaction between the three parameters. Then, using the idea of orthogonal experiment, the simulation results of the control model are analyzed by taking the maximum noise reduction as the evaluation index. It is found that the convergence coefficient has the greatest effect on the maximum noise reduction, followed by the filter order, and the reference signal gain has the least effect.

It is considered that slow convergence speed and large calculation amount of commonly used adaptive algorithm in the active control system for vehicle interior noise yield noise reduction performance and hardware requirements problems. In this paper, a 4-channel active control of vehicle interior noise based on adaptive notch filter is established, and road test is carried out to test and analyze the performance of the control system. Firstly, the general mathematic model of the multi-channel active control system based on adaptive notch filter is established. The computational complexity of the algorithm is analyzed and compared with that of the FXLMS algorithm. Secondly, a hardware-in-the-loop test bench based on multi-channel adaptive notch filter is set up, to measure the noise reduction performance of ANC system under various working conditions.

In order to realize driving safety in highway scenarios, autonomous vehicles need to predict and reason about the driving intentions and motion trajectories of surrounding target vehicles in the near feature. Essentially, trajectory prediction of target vehicles can be viewed as a typical time series generation problem, which predicts the future trajectory of the vehicle through analyzing the input of historical trajectory information or its control signals. In actual traffic scenarios, the movement between vehicles is a process of mutual game and cooperation, namely the future trajectory of a vehicle is not only related to its own historical trajectory, but also to surrounding vehicles motion. However, different surrounding traffic participants have different influence on the target vehicle, and the future motion of the vehicle is often affected by some specific surrounding traffic agents deeply.

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.

Safety is the cornerstone for Advanced Driver Assistance Systems (ADAS) and Autonomous Driving Systems (ADS). To assess the safety of a traffic situation, it is essential to predict motion states of traffic participants in the future with mathematic models. Accurate vehicle trajectory prediction is an important prerequisite for reasonable traffic situation risk assessment and appropriate decision making. Vehicle trajectory prediction methods can be generally divided into motion model based methods and maneuver model based methods. Vehicle trajectory prediction based on motion models can be accurate and reliable only in the short term. While vehicle trajectory prediction based on maneuver models present more satisfactory performance in the long term, these maneuver models rely on machine learning methods. Abundant data should be collected to train the maneuver recognition model, which increases complexity and lowers real-time performance.

Powertrain system of series-parallel hybrid vehicle contains multiple excitation sources like engine, motor and generator. The reduction of noise and vibration is quite difficult during multiplex working modes or the switch of modes. Aiming at Series-parallel hybrid powertrain system which contains engine, motor and planetary gear subsystems, this paper considered a typical working condition which is based on the power control strategy and established the torsional vibration mechanical model of hybrid powertrain system. The inherent characteristics and transient vibration response of the electric mode, hybrid mode and parking charging mode were studied and it was discovered that the repetitive frequency of the powertrain system under the three working modes is the same which is only related to inertia and meshing stiffness of planetary gear system. The non-repetitive frequency and corresponding vibration modes under the electric mode and parking charging mode are both close.