Search

Deep learning based on labeled data has brought massive success in computer vision, speech recognition, and natural language processing. Nevertheless, labeled data is just a drop in the ocean compared with unlabeled data. How can people utilize the unlabeled data effectively? Research has focused on unsupervised and semi-supervised learning to solve such a problem. Some theoretical and empirical studies have proved that unlabeled data can help boost the generalization ability and robustness under adversarial attacks. However, current theoretical research on the relationship between robustness and unlabeled data limits its scope to toy datasets. Meanwhile, the visual models in autonomous driving need a significant improvement in robustness to guarantee security and safety. This paper proposes a semi-supervised learning framework for object detection in autonomous vehicles, improving the robustness with unlabeled data. Firstly, we build a baseline with the transfer learning of an unsupervised contrastive learning method—Momentum Contrast (MoCo). Secondly, we propose a semi-supervised co-training method to label the unlabeled data for retraining, which improves generalization on the autonomous driving dataset. Thirdly, we apply the unsupervised Bounding Box data augmentation (BBAug) method based on a search algorithm, which uses reinforcement learning to improve the robustness of object detection for autonomous driving. We present an empirical study on the KITTI dataset with diverse adversarial attack methods. Our proposed method realizes the state-of-the-art generalization and robustness under white-box attacks (DPatch and Contextual Patch) and black-box attacks (Gaussian noise, Rain, Fog, and so on). Our proposed method and empirical study show that using more unlabeled data benefits the robustness of perception systems in autonomous driving.

Recent advancements in marine technology have highlighted the urgent need for enhanced underwater acoustic applications, from sonar detection to communication and noise cancellation, driving the pursuit of innovative transducer technologies. In this paper, a new underwater thermoacoustic (TA) transducer made from carbon nanotube (CNT) sponge is designed to achieve wide bandwidth, high energy conversion efficiency, simple structure, good transient response, and stable sound response, utilizing the TA effect through electro-thermal modulation. The transducer has potential application in underwater acoustic communication. An electro-thermal-acoustic coupled simulation for the open model, sandwich model, and encapsulated model is presented to analyze the transient behaviors of CNT sponge TA transducers in liquid environments. The effects of key design parameters on the acoustic performances of both systems are revealed. The results demonstrate that a short pulse excitation with a low duty cycle could greatly improve the heat dissipation of the encapsulated transducer, especially when the thermoacoustic response time becomes comparable to thermal relaxation time. [ABSTRACT FROM AUTHOR]

Vapor spontaneously diffuses from areas of high concentration to areas of low concentration as a consequence of the Second Law of Thermodynamics and the hunt for maximum entropy. The distribution of substances with lower saturated vapor pressures on a surface can create this concentration slope in the space. However, most substances, except ice, cannot maintain this slope because humidity sinking causes their saturated vapor pressure to drop. Moreover, the location of ice growth is difficult to control. In this paper, a pattern of CaCl2 was made on the micronano hierarchically aluminum substrate. When the temperature drops below zero, the vapor pressure at the position where CaCl2 is located is low relative to that of water as a way to achieve a continuous phase transition from solid CaCl2 to brine to ice. Thus, vapor nearby continuously diffuses to and deposits at the site, resulting in a lasting dry region around without water condensation. The frost growing on the pattern is also fluffy and can be easily removed by gravity and/or wind. This work has promising applications in areas where it is extremely difficult to use active deicing methods. [ABSTRACT FROM AUTHOR]

It is of great interest for autonomous vehicles to predict the trajectory of other vehicles when planning a safe trajectory. To accurately predict the trajectory of the target vehicle, the interaction between vehicles must be considered. Interaction aware prediction methods track the previous trajectories of both the target vehicle and its surrounding vehicles. In this study, the authors consider trajectory prediction as a sequence‐to‐sequence prediction problem. They tackle this problem with an LSTM encoder–decoder framework. Moreover, they propose two spatial‐attention mechanisms to account for the interaction between vehicles, i.e. context attention and lane attention. Spatial‐attention mechanisms adopt the selective‐attention mechanism of human drivers. They choose context vectors to help the model understand the surrounding environment better and thus improve its prediction accuracy. They evaluate the authors' methods on the highD data set recorded in German highways with root mean squared error metric. Their experimental results show superior performance to other state‐of‐the‐art methods. Code is available at https://github.com/momo1986/Spatial‐attention. [ABSTRACT FROM AUTHOR]

Abstract Different soil humus fractions are structurally distinct from each other molecularly, however, the relationship between their microscopic molecular structures and the macroscopic reduction of Cr(VI) is still unknown, especially for the humin fraction. In this study, different humus fractions (HA, humic acid; HMi, humin linked to iron oxides; HMc, humin linked to clay; and HMr, humin residue) were sequentially extracted from a typical black soil and well characterized. It was found that HA, HMi and HMc were the same type of humus with similar molecular structures, while HMr was structurally different from the other fractions with a high cellulose content. The removal rate of Cr(VI) in solution decreased with progressive humus fractionation, namely, HA > HMi > HMc > HMr. Based on the two-dimensional correlation spectroscopic analysis (2DCOS) of the FTIR data, the changing functional groups of all humus fractions during reacting with Cr(VI) followed a similar order: carboxyl > phenol > hydroxyl > methyl > methylene. According to the correlation analysis, Cr(VI) reduction rates by different humus fractions were mainly determined by the content of phenol (R 2 = 0.99) instead of carboxyl (R 2 = 0.28). Except for HMr, the Cr(VI) reduction rates of different humus fractions were also positively correlated with surface and bulk polarity (R 2 = 0.98 and 0.99) but not with aromaticity or aliphaticity (R 2 = 0.21). Graphical abstract Unlabelled Image Highlights • HMr is structurally different from other humus fractions with a high content of cellulose. • Cr(VI) removal rate decreased with progressively humus fractionation, namely HA > HMi > HMc > HMr. • Cr(VI) reduction rates are determined by phenol content of humus instead of carboxyl. • Cr(VI) reduction rates are also positively correlated with polarity, except for HMr. [ABSTRACT FROM AUTHOR]

Undissolved humic acid (HA) has a great retention effect on the migration of hexavalent chromium [Cr(VI)] in soil, and HA functional groups play a predominant role in this process. However, the coupled mode between Cr(VI) retention and HA functional groups reaction is still unclear. In this study, it was found that a fair amount of Cr on HA existed in the forms of ion exchangeable and binding Cr(VI) during the reaction resulting from the ion exchange adsorption and complexation of Cr(VI). According to the results of two-dimensional correlation spectroscopic analysis (2DCOS), HA functional groups participated in the reaction with Cr(VI) in the order of carboxyl ≈ chelated carboxyl > phenol > polysaccharide > methyl, and all the functional groups were more likely to be located at aromatic domains. Based on the results of XPS spectra, rather than to be oxidized by Cr(VI), carboxyl more tended to be complexed by chromium, which is regarded as the precondition for Cr(VI) reduction. Phenol, polysaccharide and methyl with distinct reaction activities successively acted as major electron donors for Cr(VI) reduction in different reaction stages. Consequently, it was determined that the retention of Cr(VI) by undissolved HA followed an adsorption-complexation-reduction mechanism, and based on this, a multi-step kinetic model with multiple types of complexation/reduction sites was developed to simulate the retention processes resulting in a much better fitting effect ( R 2 > 0.99) compared with traditional first-order and second-order kinetic models ( R 2 < 0.95). This demonstrated that the multi-step kinetic model is of great potential in accurately simulating the migration and transformation of Cr(VI) in soil environment. [ABSTRACT FROM AUTHOR]

Undissolved humic acid (HA) is known to substantially effect the migration and transformation of hexavalent chromium [Cr(VI)] in soils. The mechanisms of Cr(VI) retention in soils by undissolved HA have been reported; however, past studies are inconclusive about the types of HA functional groups that are involved in Cr(VI) retention and the retention mechanisms. Utilizing a two-dimensional correlation spectroscopy (2DCOS) analysis for FTIR and 13 C CP/MAS NMR, this study investigated the variations of HA function groups and molecular structures after reactions with aqueous Cr(VI) under different pH conditions. Based on the changing sequence of functional groups interpreted from the 2DCOS results, a four-step mechanism for Cr(VI) retention was determined as follows: (1) electrostatic adsorption of Cr(VI) to HA surface, (2) complexation of adsorbed Cr(VI) by carboxyl and ester, (3) reduction of complexed Cr(VI) to Cr(III) by phenol and polysaccharide, and (4) complexation of reduced Cr(III) by carboxylic groups. These functional groups that are involved in Cr(VI) retention were determined to occur in aromatic domains. [ABSTRACT FROM AUTHOR]

Abstract-This paper proposes new methods to extract ground points and segment non-ground objects based on the spherical coordinates of point clouds collected by the 3-D LiDAR sensor system. Ground points extraction and non-ground objects segmentation are two primary issues in environment perception of autonomous driving. Firstly, the new method to extract ground points is described in this paper. It is based on the breakpoints and turning points of the radial distance curve in the spherical coordinates without assuming the road is located in the lowest. Then the algorithm for non-ground objects segmentation is proposed, which uses azimuth angle and radial distance in the spherical coordinates as judgment criterion. This algorithm works efficiently and it can avoid over-segmentation and under-segmentation. Finally, the experimental results using the point clouds acquired by the sensor Velodyne HDL-32E are presented. Comparing to the existing methods, the results show the advantages of methods based on the spherical coordinates. [ABSTRACT FROM PUBLISHER]

Imperfect preventive maintenance (PM) activities are very common in industrial systems. For condition-based maintenance (CBM), it is necessary to model the failure likelihood of systems subject to imperfect PM activities. In this paper, the models in the field of survival analysis are introduced into CBM. Namely, the generalized accelerated failure time (AFT) frailty model is investigated to model the failure likelihood of industrial systems. Further, on the basis of the traditional maximum likelihood (ML) estimation and expectation maximization (EM) algorithm, the hybrid ML-EM algorithm is investigated for the estimation of parameters. The hybrid iterative estimation procedure is analyzed in detail. In the evaluation experiment, the generated data of a typical degradation model are verified to be appropriate for the real industrial processes with imperfect PM activities. The estimates of the model parameters are calculated using the training data. Then, the performance of the model is analyzed through the prediction of remaining useful life (RUL) using the testing data. Finally, comparison between the results of the proposed model and the existing model verifies the effectiveness of the generalized AFT frailty model. [ABSTRACT FROM AUTHOR]

In the present work, three NiTiNb alloys with nominal compositions of Ni50–x/2Ti50–x/2Nbx (x = 18, 20 and 22, at.%) were prepared. The microstructure, thermal-induced phase transformation, and phase components of NiTiNb hypoeutectic, eutectic, and hypereutectic alloys were investigated by scanning electron microscopy coupled with energy dispersive spectrometer (SEM/EDS), differential scanning calorimetry (DSC), and X-ray diffraction (XRD), respectively. Two peaks occurred in the DSC curves of Ni41Ti41Nb18 solution at 750 and 850 °C. The different microstructures of NiTiNb hypoeutectic, eutectic and hypereutectic alloys, the occurrence of two peaks in the DSC curves, and the characteristic of martensite transformation temperature were analyzed and discussed. The results show that different Nb content leads to different microstructures of Ni50–x/2Ti50–x/2-Nbx alloys. Moreover, the difference of Ni/Ti ratio between the primary NiTi matrix and NiTi in eutectic structure strongly influenced the phase transformation behavior that the DSC curves of Ni41Ti41Nb18 alloy show two peaks. Furthermore, under the same solution treatment, the Ni40Ti40Nb20 alloy has a lower Ms than the other two alloys. The Ms has a tendency to decrease with an increase in solution temperature. [ABSTRACT FROM AUTHOR]

Abstract The retention of Cr(VI) in subsurface environment is highly dependent on humic acid (HA), however, the undissolved form is poorly investigated, the amount of which can be of two magnitude higher compared with the dissolved one in soils and sediments. In this study, the effects of time, initial concentration, pH, ionic strength, ion species and temperature on the adsorption and reduction respective processes by undissolved self-extracted peat soil HA from Northeast China (EHA) and from Sigma Aldrich (CHA) were investigated by batch experiments. Cr(VI) removal rates by EHA were higher than CHA and the maximum Cr(VI) removal amount for EHA and CHA were 0.77 (±0.01) and 0.61 (±0.02) mmol/g. Of these, 98% and 54% were reduced to Cr(III) by EHA and CHA respectively, which were related to the phenolic group content of HA. With time, the adsorbed Cr(VI) on HA increased to a maximum level (equilibrium) beyond which Cr(VI) reduction dominated the removal process. Cr(VI) adsorption and reduction by undissolved HA increased as pH decreased. Co-existing ion species had varying effect on Cr(VI) adsorption and indirectly on reduction especially divalent cations which was suggestive of cation bridging between Cr anions and ionized carboxyl group of HA. The positive effect of ionic strength (Ca2+) on Cr(VI) adsorption through complexation corroborated the cation bridge effect of divalent cations. Temperature increased both Cr(VI) adsorption (complexation) and reduction with enhancing reduction rate constants and partitioning. ΔH o , ΔS o , and ΔG⁰ parameters showed that Cr(VI) adsorption and reduction processes were endothermic, irreversible and spontaneous. Highlights • Verified the adsorption-reduction mechanism under various conditions. • Established the relationship between adsorption and reduction under varied conditions. • The slope of the relationship suggested different reduction rate constants. • Ca2+ functioned as a cation bridge enhancing adsorption and indirectly reduction. • Temperature enhanced adsorption (distribution coefficient) and reduction rate constant. [ABSTRACT FROM AUTHOR]

Undissolved humic acid (HA) is known to substantially effect the migration and transformation of hexavalent chromium [Cr(VI)] in soils. The mechanisms of Cr(VI) retention in soils by undissolved HA have been reported; however, past studies are inconclusive about the types of HA functional groups that are involved in Cr(VI) retention and the retention mechanisms. Utilizing a two-dimensional correlation spectroscopy (2DCOS) analysis for FTIR and 13 C CP/MAS NMR, this study investigated the variations of HA function groups and molecular structures after reactions with aqueous Cr(VI) under different pH conditions. Based on the changing sequence of functional groups interpreted from the 2DCOS results, a four-step mechanism for Cr(VI) retention was determined as follows: (1) electrostatic adsorption of Cr(VI) to HA surface, (2) complexation of adsorbed Cr(VI) by carboxyl and ester, (3) reduction of complexed Cr(VI) to Cr(III) by phenol and polysaccharide, and (4) complexation of reduced Cr(III) by carboxylic groups. These functional groups that are involved in Cr(VI) retention were determined to occur in aromatic domains., (Copyright © 2017 Elsevier Ltd. All rights reserved.)