Search

Aiming at the problem that the existing motion planning methods for autonomous driving ignore the interaction of surrounding vehicles when extracting state information and the bad planning effect in dense traffic scenarios, a motion planning model combined with graph neural network and deep reinforcement learning is proposed. Firstly, based on the graph neural network, an interactive feature representation method of self-driving vehicles is proposed to extract spatial interaction features of multiple traffic participants. In this case, a learning strategy for motion planning is designed based on twin delayed deep deterministic policy gradient (TD3), and the next action is predicted from the interactive features so as to realize motion planning. The proposed method is compared with the current motion planning model LSTM+TD3, TD3 and deep deterministic policy gradient (DDPG) for autonomous driving, in dense traffic scenarios, the experimental results of training and testing in the PGDrive driving simulator increased by 36%, 43%, 23% and 13, 19, 53 percentage points compared with the comparison method, which means the proposed method can effectively solve the problem of interactive information perception of surrounding vehicles for better motion planning of autonomous driving. [ABSTRACT FROM AUTHOR]

Two-dimensional carbon nitrides (CxNy) have gained significant attention in various fields including hydrogen energy development, environmental remediation, optoelectronic devices, and energy storage owing to their extensive surface area, abundant raw materials, high chemical stability, and distinctive physical and chemical characteristics. One effective approach to address the challenges of limited visible light utilization and elevated carrier recombination rates is to establish heterojunctions for CxNy-based single materials (e.g. C2N3, g-C3N4, C3N4, C4N3, C2N, and C3N). The carrier generation, migration, and recombination of heterojunctions with different band alignments have been analyzed starting from the application of CxNy with metal oxides, transition metal sulfides (selenides), conductive carbon, and Cx′Ny′ heterojunctions. Additionally, we have explored diverse strategies to enhance heterojunction performance from the perspective of carrier dynamics. In conclusion, we present some overarching observations and insights into the challenges and opportunities associated with the development of advanced CxNy-based heterojunctions. [ABSTRACT FROM AUTHOR]

We conducted high-throughput calculations on the structure library of two-dimensional covalent organic frameworks (2D COFs), with the aim of finding new candidate materials as electron transport layers (ETLs) for the high-performance perovskite light-emitting diodes (PeLEDs). Starting with nearly 400 unique 2D COFs with hexagonal (hcb) topology, 23 candidates for β-ketoenamine linked 2D COFs composed of 1,3,5-triformylphloroglucinol (Tp) and linear diamines are identified by band structures. Interestingly, 2 β-ketoenamine linked 2D Tp-DAAQ and 2D Tp-DABDA with electron-accepting substituents, are the most promising candidates due to the optimal band edge and effective mass. Further, the nonpolarity of the Tp-DAAQ molecule makes it a more favorable choice for the ETL, as confirmed by nonadiabatic molecular dynamics that 2D Tp-DAAQ with anthraquinone diamine linker has a longer e–h nonradiative recombination time. This work provides a paradigm for the design of high-performance ETLs. [ABSTRACT FROM AUTHOR]

Benefiting from the unique electronic and structural characteristics, transition metal and nitrogen co-doped carbon-based (M–N–C) nanomaterials are considered to be one of the most promising catalysts. The combination of uniformly dispersed Cu sites and N-doped carbon materials has enabled Cu–N–C SACs to show outstanding performances in many catalytic fields. However, the adjustment of their coordination environment, the influence of carrier materials and their dynamic evolution in the catalytic process still remain unclear, which limits the application of Cu–N–C SACs. Thus, it is of great significance and urgent to deeply understand the reaction mechanism of Cu–N–C SACs in different catalytic fields. This review focuses on the preparation strategy, regulation mechanism of the coordination environment, catalytic performance and dynamic evolution process of Cu–N–C SACs. Combined with a series of in situ/operando characterization methods, the changes in their coordination and structure in the catalytic process is highlighted. Finally, the prospects of active machine learning, coordination structure identification and determination of real active sites are presented. This review aims to provide a reference for the rational design of Cu–N–C SACs and the exploration of their catalytic mechanism. [ABSTRACT FROM AUTHOR]

Millions of families around the world remain vulnerable to water scarcity and have no access to drinking water. Advanced oxidation processes (AOPs) are an effective way towards water purification with qualified reactive oxygen species (ROSs) while are impeded by the high-cost and tedious process in either input of consumable reagent, production of ROSs, and the pre-treatment of supporting electrolyte. Herein, we couple solar light-assisted H2O2 production from water and photo-Fenton-like reactions into a self-cyclable system by using an artificial leaf, achieving an unassisted H2O2 production rate of 0.77 μmol/(min·cm2) under 1 Sun AM 1.5 illumination. Furthermore, a large (70 cm2) artificial leaf was used for an unassisted solar-driven bicarbonate-activated hydrogen peroxide (BAP) system with recycled catalysts for real-time wastewater purification with requirements for only water, oxygen and sunlight. This demonstration highlights the feasibility and scalability of photoelectrochemical technology for decentralized environmental governance applications from laboratory benchtops to industry. Continuous generation of reactive oxygen species is desirable in the advanced oxidation process. Here, the authors report a self-cycled photoFenton-like with a scalable artificial leaf for production of H2O2 from water with solar-to-H2O2 efficiency of 1.46%. [ABSTRACT FROM AUTHOR]

A growth trial was conducted to evaluate the effects of dietary lipid levels on the lipid deposition and metabolism of subadult triploid rainbow trout. Diets with low (148 g lipid kg-1 diet), moderate (228 g lipid kg-1 diet), and high (294 g lipid kg-1 diet) crude lipid contents were fed to quadruplicated groups of fish (233 g ± 0.2 g) for 80 days, and they were named as LL, ML, and HL, respectively. Results showed that the lowest and highest values of condition factor and hepatosomatic index were shown in the LL group, respectively, while the HL group obtained the highest liver redness value, plasma total cholesterol, high-density lipoprotein-cholesterol, and low-density lipoprotein-cholesterol contents (P < 0.05). The viscerosomatic index and plasm glucose, nonesterified fatty acids, and triglyceride contents were comparable among groups (P > 0.05). As for lipid deposition, viscera and muscle were the main lipid storage place for triploid rainbow trout when tissues' weight is taken into consideration. Overall quantitative PCR showed that the lipid uptake, triglyceride, and fatty acid catabolism were upregulated, but glycolysis was downregulated as dietary lipid increased. The expression of genes involved in lipogenesis, lipoprotein clearance, very low-density lipoprotein assembly, and glycogenesis was not affected by dietary lipid levels (P > 0.05). In summary, subadult triploid rainbow trout store lipid mainly in viscera and muscle; it could maintain hepatic lipid homeostasis when fed with dietary lipid levels ranging from 150 to 300 g kg-1 by regulating lipid uptake and switching energy supply between glycolysis and fatty acids β-oxidation. [ABSTRACT FROM AUTHOR]

The deterioration of water ecology caused by the discharge of oil spill wastewater, industrial sewage, and municipal sewage has attracted wide attention worldwide. Thus, it is significant to design a simple, environmentally friendly approach to separate oil–water mixtures. In this work, three different fabrics with pH-induced wettability transition were prepared by a dip-coating process for oil and water separation. The dip-coating fabrics had the advantages of oil–water separation, photocatalytic degradation, and recycling. Polyethylene terephthalate/polyamide nonwoven fabric was used as the substrate materials of the fabric. The carboxylic acid-modified TiO2 endowed the fabric with hydrophilicity–hydrophobicity and photocatalytic properties. The Fe3O4 nanoparticles obtained by the coprecipitation method provided magnetism for the fabric, facilitating the recycling of the fabric and improving the hydrophobicity of the fabric. The fabrics coated with dipping solutions were superhydrophobic in a neutral environment and hydrophilic in an alkaline environment. Among the three coated fabrics, the fabric coated with stearic acid/TiO2-Fe3O4 (FST) had the most satisfying oil–water separation performance and durability. Under the neutral condition, the contact angle of the FST was 151° and the separation efficiency was 98%. Under the alkaline condition, the underwater oil contact angle of the FST was 150° and the separation efficiency was 95%. After 15 cycles, the oil–water separation rate of the FST was still higher than 90%. Due to the presence of TiO2, the coated fabric had an exceptional performance in the photodegradation of organic pollutants (69.9%). In addition, the fabrics can be quickly recovered due to magnetism. [ABSTRACT FROM AUTHOR]

Acute acquired comitant esotropia (AACE) is characterised by the acute onset of diplopia which affects work and life. To describe the clinical characteristics and discuss the aetiology of acute acquired comitant esotropia. The medical records of 51 patients with AACE were retrospectively analysed, and their age, occupation, daily average time spent on near work, angle of deviation, refractive error and aetiology were analysed. The age at onset of AACE was eight to 55 years. Thirty-nine patients were aged 12 to 36 years (76.5%), and 42 patients (82.4%) were myopic. The angle of deviation was 20Δ (12Δ to 35Δ) during near fixation and 25Δ (18Δ to 40Δ) during distance fixation. The deviation during distance fixation was larger than that during near fixation, the difference being statistically significant (p < 0.001). The daily average time of near work before onset was 9.0 h (8.0 h to 10.0 h); 35 patients (68.6%) spent more than 8 h performing near work. Their occupations included students, accountants, information technology staff, and those who required electronic use for a long time. There is no correlation between the angle of deviation and the time of near work (p > 0.05). Among the 51 patients, three had intracranial diseases. Diplopia resolved in 47 cases following treatment. The angle of deviation post-treatment was 0Δ (0Δ to 5Δ) during near and 0Δ (0Δ to 3Δ) during distance fixation. AACE occurs mostly in myopic older children and adults. The onset of esotropia is related to long-term near work. AACE patients mostly present an angle of deviation that is larger during distance than near fixation. Intracranial disease is an infrequent but important cause of AACE that needs to be excluded in all cases. [ABSTRACT FROM AUTHOR]

Motion planning is one of the most significant part in autonomous driving. Learning-based motion planning methods attract many researchers’ attention due to the abilities of learning from the environment and directly making decisions from the perception. The deep Q-network, as a popular reinforcement learning method, has achieved great progress in autonomous driving, but these methods seldom use the global path information to handle the issue of directional planning such as making a turning at an intersection since the agent usually learns driving strategies only by the designed reward function, which is difficult to adapt to the driving scenarios of urban roads. Moreover, different motion commands such as the steering wheel and accelerator are associated with each other from classic Q-networks, which easily leads to an unstable prediction of the motion commands since they are independently controlled in a practical driving system. In this paper, a conditional deep Q-network for directional planning is proposed and applied in end-to-end autonomous driving, where the global path is used to guide the vehicle to drive from the origination to the destination. To handle the dependency of different motion commands in Q-networks, we take use of the idea of fuzzy control and develop a defuzzification method to improve the stability of predicting the values of different motion commands. We conduct comprehensive experiments in the CARLA simulator and compare our method with the state-of-the-art methods. Experimental results demonstrate the proposed method achieves better learning performance and driving stability performance than other methods. [ABSTRACT FROM AUTHOR]

In autonomous driving, many learning-based methods for motion planing have been proposed in literature, which can predict motion commands directly from the sensory data of the environment, but these methods can neither predict multiple motion commands, such as steering angle, accelerator and brake, nor balance errors among different motion commands. In this paper, we propose a deep cascaded neural network for predicting multiple motion commands which can be trained in an end-to-end manner for autonomous driving. The proposed deep cascaded neural network consists of a convolutional neural network (CNN) and three long short-term memory (LSTM) units, fed with images from a front-facing camera installed at the vehicle. As the outputs, the proposed model can predict thee motion planning commands simultaneously including steering angle, acceleration, and brake to enable the autonomous driving. In order to balance errors among different motion commands and improve prediction accuracy, we propose a new network training algorithm, where three independent loss functions are designed to separately update the weights in the three LSTMs connected to three motion commands. We conduct comprehensive experiments using the data from a driving simulator and compare our method with the state-of-the-art methods. Simulation results demonstrate the proposed motion planning model achieves better accuracy performance than other models. [ABSTRACT FROM AUTHOR]

The electrical contacts at the van der Waals (vdW) interface between two-dimensional (2D) semiconductors and metal electrodes could dramatically affect the device performance. Herein, we construct a series of graphene (Gr)/XC (X = P, As, Sb, and Bi) vdW heterostructures, in which XC monolayers have aroused considerable attention recently as an emerging class of 2D semiconductors. The electronic structures and contact properties of Gr/XC vdW heterostructures are investigated systematically using first-principles calculations. The band structures indicate that both Gr/PC and Gr/AsC heterostructures form n-type Schottky contacts with Schottky barrier heights (SBHs) of 0.01 eV and 0.43 eV, respectively, while both Gr/SbC and Gr/BiC heterostructures preferably form Ohmic contacts. The different X atoms result in different work functions, electron flows, charge distributions and orientations of the dipole moment in Gr/XC heterostructures. Moreover, the tunneling probabilities increase with the increasing atom radius of X from P to Bi, indicating the most improved current and smaller contact resistance at the interfaces of Gr/BiC compared to Gr/PC, Gr/AsC and Gr/SbC heterostructures. Our work could provide meaningful information for designing high-performance nanoelectronic devices based on Gr/XC heterostructures. [ABSTRACT FROM AUTHOR]

Reactive dyes are the main dyes in printing and dyeing of cellulosic fibers. Reactive dyes fixation rate is a vital indicator to measure the degree of the covalent bond between cellulose and reactive dyes. However, the determination of the fixation rate is tedious and time-consumptive. Based on the theory of reactive dyes dyeing and application of modern computer color matching technology, the relationship between K/S value and the fixation rate with the reactive dyes on cotton fabric was studied. The feasibility of K/S value instead of the traditional washing method for the determination of reactive dyes fixation rate was proved. In this study, the K/S value of the fabric has an excellent linear relationship to the reactive dyes fixation rate obtained by the washing method. The reactive dyes fixation rate can be obtained through the K/S correction value. [ABSTRACT FROM AUTHOR]

One of the challenges with supercapacitors is achieving ultrahigh specific capacitance by maximizing utilization of active materials while assuring their cycling stability. Herein, we report a significant advance toward the design and fabrication of novel hierarchical hexagonal prism arrays constructed using carbon doped mixed-valence Co–Mn–Fe multi-metallic phosphide ultrathin porous nanoflakes via an ion-induced growth of the precursor followed by phosphorization. Each hexagonal prism features an architecture consisting of six symmetrically distributed and interlaced nanoflakes each with a thickness of ∼15 nm and many pores of ∼10 nm diameter. New insights into the Fe2+ induced anisotropic growth of Co–Mn oxide hydroxides and partial conversion of Co2+ into Co3+ leading to enhanced electrical conductivity are proposed. Such a Co–Fe–Mn phosphide electrode shows a specific areal capacitance reaching 4.36 F cm−2 (at 2 A cm−2), much higher than those for bimetallic phosphides. The high areal capacitance is attributed to the unique architecture lowering flow resistance of the electrolyte wetting process, abundant mixed-valence states of metals and active sites/interfaces, and enhanced electrical conductivity. The carbon doping substantially improves the cycling stability because it effectively buffers the strain/volume expansion during fast faradaic reactions. Especially even at a high current density of 20 mA cm−2, 100% capacitance retention can be achieved after 5000 cycles. An asymmetric device based on such an electrode delivers an energy density of 53.2 W h kg−1 (at 399.7 W kg−1) which is higher than many previously reported values. [ABSTRACT FROM AUTHOR]

To handle the issue of preventing emergencies for motion planning in autonomous driving, we present a novel parallel motion planning framework. Artificial traffic scenes are firstly constructed based on real traffic scenes. A deep planning model which can learn from both real and artificial scenes is developed and used to make planning decisions in an end-to-end mode. To prevent emergencies, a generative adversarial networks (GAN) model is designed and learns from the artificial emergencies from artificial traffic scenes. During deployment, the well-trained GAN model is used to generate multiple virtual emergencies based on the current real scene, and the well-trained planning model simultaneously makes different planning decisions for both virtual scenes and the current scenes. The final planning decision is made by comprehensively analyzing observations and virtual emergencies. Through parallel planning, the planner can timely make rational decision without a large number of calculations when an emergency occurs. [ABSTRACT FROM AUTHOR]