Search

Using deep reinforcement learning to solve traffic signal control problems is a research hotspot in the intelligent transportation field. Researchers have recently proposed various solutions based on deep reinforcement learning methods for intelligent transportation problems. However, most signal control optimization takes the maximization of traffic capacity as the optimization goal, ignoring the concerns of pedestrians at intersections. To address this issue, we propose a pedestrian-considered deep reinforcement learning traffic signal control method. The method combines a reinforcement learning network and traffic signal control strategy with traffic efficiency and safety aspects. At the same time, the waiting time of pedestrians and vehicles passing through the intersection is considered, and the Discrete Traffic State Encoding (DTSE) method is applied and improved to define the more comprehensive states and rewards. In the training of the neural network, the multi-process operation method is adopted, and multiple environments are run for training simultaneously to improve the model’s training efficiency. Finally, extensive simulation experiments are conducted on actual intersection scenarios using the simulation software Simulation of Urban Mobility (SUMO). The results show that compared to Dueling DQN, the waiting time due to our method decreased by 58.76% and the number of people waiting decreased by 51.54%. The proposed method can reduce both the number of people waiting and the waiting time at intersections.

Vertically oriented graphene (VG) has attracted attention for years, but the growth mechanism is still not fully revealed. The electric field may play a role, but the direct evidence and exactly what role it plays remains unclear. Here, we conduct a systematic study and find that in plasma-enhanced chemical vapor deposition, the VG growth preferably occurs at spots where the local field is stronger, for example, at GaN nanowire tips. On almost round-shaped nanoparticles, instead of being perpendicular to the substrate, the VG grows along the field direction, that is, perpendicular to the particles' local surfaces. Even more convincingly, the sheath field is screened to different degrees, and a direct correlation between the field strength and the VG growth is observed. Numerical calculation suggests that during the growth, the field helps accumulate charges on graphene, which eventually changes the cohesive graphene layers into separate three-dimensional VG flakes. Furthermore, the field helps attract charged precursors to places sticking out from the substrate and makes them even sharper and turn into VG. Finally, we demonstrate that the VG-covered nanoparticles are benign to human blood leukocytes and could be considered for drug delivery. Our research may serve as a starting point for further vertical two-dimensional material growth mechanism studies.

Building extraction from high-resolution remote sensing images is of great significance in urban planning, population statistics, and economic forecast. However, automatic building extraction from high-resolution remote sensing images remains challenging. On the one hand, the extraction results of buildings are partially missing and incomplete due to the variation of hue and texture within a building, especially when the building size is large. On the other hand, the building footprint extraction of buildings with complex shapes is often inaccurate. To this end, we propose a new deep learning network, termed Building Residual Refine Network (BRRNet), for accurate and complete building extraction. BRRNet consists of such two parts as the prediction module and the residual refinement module. The prediction module based on an encoder–decoder structure introduces atrous convolution of different dilation rates to extract more global features, by gradually increasing the receptive field during feature extraction. When the prediction module outputs the preliminary building extraction results of the input image, the residual refinement module takes the output of the prediction module as an input. It further refines the residual between the result of the prediction module and the real result, thus improving the accuracy of building extraction. In addition, we use Dice loss as the loss function during training, which effectively alleviates the problem of data imbalance and further improves the accuracy of building extraction. The experimental results on Massachusetts Building Dataset show that our method outperforms other five state-of-the-art methods in terms of the integrity of buildings and the accuracy of complex building footprints.

In the present study, the ZnO-porous carbon (ZnO-C) composites were prepared by pyrolyzing MOF-74 (Zn) precursor at different pyrolysis temperatures. The ZnO-C composites were endowed with ultrafast organic dye adsorption capability and promising supercapacitance properties due to the existence of abundant pores within the composite structures. Having a surface area of 782.971 m2/g and pore volume of 0.698 m3/g, the composite pyrolyzed at 1000 °C (ZnO-C1000) exhibited the best performance for organic pollutant uptake. Specifically, 50 mg of ZnO-C1000 could remove all the Rhodamine B dye from 100 mL aqueous solution within 0.5 h even the dye concentration was as high as 40 mg/L. It was also shown that the ZnO-C composites could preserve their adsorption capability in a wide pH range, and keep promising dye adsorption stability after consecutive adsorption/desorption cycles. Furthermore, the ZnO-C900 exhibited a specific capacitance of 197.84 F/g as the supercapacitance electrode with good stability (∼97.8% capacitance retention after 1000 cycles). The overall results indicate that the prepared ZnO-C composites have multi-application potentials which can be utilized as efficient pollutant absorbents as well as electrode materials for supercapacitors.

A binder free electrode is fabricated by “planting” Anderson type polyoxometalate derived NiS/MoS2 “follower” on functional carbon paper. This electrode shows excellent UOR activity. In 1 ​M KOH containing 0.4 ​M urea, it only needs 1.42 and 1.43 ​V voltage to achieve 50 and 100 ​mA ​cm−2 current. On the contrary, to reach 50 and 100 ​mA ​cm−2, OER require 1.68 and 1.75 ​V. Besides promising UOR activity, the HER property of this electrode is also attractive. In urea containing electrolyte, its η10 and η100 are only 128 and 283 ​mV. More importantly, it shows excellent stability in both UOR and HER. In electrolytic cell, with this electrode as cathode and anode simultaneously, only 1.42 and 1.49 ​V voltage is needed to obtain 10 and 50 ​mA ​cm−2 current in urea containing electrolyte. Without urea, the voltages increase to 1.69 and 1.75 ​V. The excellent UOR and HER properties suggest this binder free electrode is competent for low energy consumption H2 production.