Search

The primary goal of the paper is to explore the human-vehicle-road interaction mechanism in the traffic environment and evaluate the traffic environment complexity for unmanned vehicles in urban roads. In particular, we propose the quantitative evaluation models of the traffic environment complexity for unmanned vehicles in urban roads in the paper. Specifically, the structure system of the complex traffic environment in urban roads is dissected from the aspect of human-vehicle-road, laying the basis for proposing influencing factors of traffic environment complexity. We divide the complex traffic environment into the static traffic environment and the dynamic traffic environment in light of relative static and dynamic characteristics of various environmental elements. For the complexity of the static traffic environment, the quantitative evaluation model is established by the grey relation analysis method that converts static environment complexity into the relation degree of static complexity’s influencing factors. For the complexity of the dynamic traffic environment, the quantitative evaluation model is established based on the improved gravitation model that introduces the concepts of equivalent mass and the contribution degree of the unmanned vehicles’ driving strategy. Besides, we evaluate the traffic environment complexity in the designed scenario by quantitative models proposed in the paper and existing evaluation models of traffic environment complexity in urban roads. The calculating process and results show that the proposed quantitative models of traffic environment complexity are more convenient and more reasonable, which provide a new idea and a method to evaluate the traffic environment complexity.

Urban geological information management systems based on a B/S structures are an inevitable trend in the development of smart geology and big data. Such systems are important platforms for promoting the sharing of data and services among different city entities. It is necessary to overcome the limitations of C/S or partial B/S structures and establish an urban geological information management system based on a comprehensive B/S structure, which is currently rare. According to the above questions, this paper focuses on three types of technical tasks to construct an urban geological information management system based on a B/S structure. In terms of data management, we leave many functions to the server, and group data by type. In terms of data application and sharing, we quickly query and graphically display the data from massive data sets, and perform collaborative analyses of different professional data. In terms of 3D geological modelling and analysis, in order to efficiently provide large-scale 3D geological model analysis in the browser, we explore a new data structure and algorithm for optimization modelling to reduce the amount of model and the number of calculations. Using the above technical methods, we develop a geological information management system for a city in eastern China. The system realizes the “One Map” function for land and resources, creates a platform for the sharing of geological data and services among different entities, and provides technical support for the construction of “smart cities”.

Nowadays, in the area of autonomous driving, the computational power of the object detectors is limited by the embedded devices and the public datasets for autonomous driving are over-idealistic. In this paper, we propose a pipeline combining both block-wise pruning and channel-wise pruning to compress the object detection model iteratively. We enforce the introduced factor of the residual blocks and the scale parameters in Batch Normalization (BN) layers to sparsity to select the less important residual blocks and channels. Moreover, a modified loss function has been proposed to remedy the class-imbalance problem. After removing the unimportant structures iteratively, we get the pruned YOLOv3 trained on our datasets which have more abundant and elaborate classes. Evaluated by our validation sets on the server, the pruned YOLOv3 saves 79.7% floating point operations (FLOPs), 93.8% parameter size, 93.8% model volume and 45.4% inference times with only 4.16% mean of average precision (mAP) loss. Evaluated on the embedded device, the pruned model operates about 13 frames per second with 4.53% mAP loss. These results show that the real-time property and accuracy of the pruned YOLOv3 can meet the needs of the embedded devices in complicated autonomous driving environments.

Combined the development trend of the logistics industry and the current problems and challenges facing the industry, this paper analyzes the needs of the logistics industry, explores the role of edge cloud synergy as an infrastructure in promoting the digital transformation of the logistics industry and studies the technical architecture of smart terminals, edge computing and cloud platform technologies in smart logistics scenarios, so as to build an open information infrastructure platform. The paper focuses on the application of five scenarios: smart transportation management, warehousing park management, smart distribution management, cloudification of logistics system, and logistics blockchain.

The Social Internet of Things(SIoT) introduces the concept of social relations to objects, so as to realize the discovery of objects and services, the interaction between objects, and the connection of goods network. The cloud-services in SIoT which are capable of perceiving real environments, can combine with indoor positioning technologies based on Wireless Networking Technologies to provide convenient services to users. However, indoor positioning algorithms using the received signal strength indicator (RSSI) are vulnerable to environmental interference, and the positioning results are unstable. This paper proposes an indoor positioning technology based on multidimensional spatial similarity. In particular, the fingerprint node database is optimizing by the Machine Learning(ML) mechanism in the cloud. The fingerprint nodes are screened by multidimensional spatial similarity. Moreover, the distance among the nodes is used as a weight factor to improve the traditional triangle localization algorithm. Furthermore, the weighted median Gaussian filter which can reduce the adverse effects of noise on the positioning accuracy is used to improve the efficiency of target positioning. Finally, tests are conducted in a laboratory environment. The results show that the accuracy of our method is higher than that of traditional indoor positioning methods, while the average positioning error is reduced significantly. In summary, our method has stronger anti-interference ability and meets the requirements of indoor positioning.

With the explosive increase of remote sensing images, content-based remote sensing image retrieval (CBRSIR) has aroused widespread attention. Convolutional Neural Network (CNN) based methods are widely used in CBRSIR due to the development of deep learning. However, common used CNN models have difficulties in holding shift-invariant property due to the widely used down-sampling method, which means a little shift of input may cause a mutation of feature representation. To mitigate the absence of shift-invariant in down-sampling, we propose the learnable blur kernel (LBK), that can enhance the feature extraction capability by leveraging more context information. We build on this concept without extra cost, which can be simply integrated with modern CNNs architecture. Our method is validated on the public remote sensing dataset and compared with other retrieval methods. The overall experimental results show that the proposed method achieves outstanding performance.

Object detection in remote sensing images is a challenging task because of the large scale variations across the geospatial objects. The feature pyramid network (FPN) is widely used to alleviate the scale variations problem, however, it only fuses the features from adjacent levels and lacks the information of the entire feature hierarchy. In this paper, we propose a novel and effective feature pyramid aggregation network, called Adaptive Feature Aggregation Network (AFANet). Specifically, we propose the Adaptive Feature Aggregation (AFA) module to adaptively aggregate multi-level features of FPN and introduce the Bottom-up Path to enhance the location information of the entire feature levels. In addition, we use the Receptive Field Block (RFB) module to capture different receptive field features for each level feature map. We evaluate the effectiveness of our AFANet on the DOTA dataset and achieves noticeable performance compared with the baseline.

The operational index evaluation needs to be combined with the static capability and dynamic tactics of arms of both sides. According to the Lanchester equation of multi-arms engagement, we propose an operational index evaluation method based on the greedy strategy. The proposed method combines the normal engagement model and the reinforcement learning theory, and circularly updates the operational index and fire assignment of arms, which gradually converges to a stable value. The method has the advantages of intuitive principle, bootstrap and convenient for computer processing.

This paper presents a 25–29 GHz 64-element small-cell (RF to IF) dual-polarized/dual-beam phased-array for 5G communication systems. The small-cell includes dual-polarized antennas, sixteen 2 × 2 beamformers, a dual-channel driver (TX/RX), a dual-channel transceiver, filters, and a local oscillator (LO) on the printed-circuit board. This small-cell provides a measured EIRPPsat of 55 dBm for each polarization with 4 GHz, and 40–140 dBc single-sideband rejection at an IF of 2–8 GHz. The TX output spectrum is measured at an IF of 5 and 6 GHz, and all the spurs are < -55 dB lower than the desired RF signal at 10-dB back-off operation. A complex modulation measurement using OFDM 5GNR 64-QAM 400 MHz waveform (PAPR = 11–12 dB) is performed and a 43–45 dBm EIRP is achieved at 4-5% EVMRMS (- 28 to -26 dB) for the small-cell. To our knowledge, this is the first demonstration of a high-performance, low-spur phased-array for 25–29 GHz 5G systems.

In order to improve the student’s ability under the Emerging Engineering Education, the Blended learning in college physics experiments has been explored. In this study, a new learning method was developed using the combination of online and face-to-face class for students in University. This new learning method is characterized by absolute autonomy in the classroom. The students could complete self e-learning through the process of the class assisted by multimedia and teachers.

Rugby is a fair compete fullfed with human discipline, self-control, and mutual respect between athletes. Notably, tackle action is one of the most basic rugby actions and the smallest unit of body touch and confrontation when attacker and defender fight for possession of rugby. After match, trainer usually count the number of tackle action happened during match to analyze and discuss for adjusting rugby tactics. This paper studies a method that optimized BP neural network based on genetic algorithm and build a neural network classifier through extracting 10-dimension gray characteristics of samples and 12-dimension geometric characteristics of body to recognize and count the tackle action. After further optimizing network structure and weights, experimental results show that our model could tackle action happened in match to avoid the subjective and empirical nature of artificial viewing discrimination with strong practical value, and will greatly improve the efficiency of trainers' work.

The ray model proposed by M.A. Alonso and M.R. Dennis gives a ray-based interpretation of well-known Gaussian mode families, and it opens the door to design new structured Gaussian beams. Based on the ray model, the wave field of octagonal structured Gaussian beam of different parameters N and µ is constructed with its expansion in the Laguerre-Gaussian beam basis which promises potential realization in fiber. In addition, the influence of parameters on the light field distribution of octagonal structured Gaussian beams is discussed.

Understanding deformations in biological systems (e.g., parietals of rhinolophus ferrumequinum) and their relations with ultrasonic functionalities is a fundamental science problem. It has promising potential of providing insightful guidance for design and manufacture of artificial sonars with promising potential. However, previous research encounters enormous real-world constraints, especially limited by environmental settings and availability of experimental animals. Moreover, existing works mainly focus on analyzing statistical correlations between geometric features and acoustic functionalities, and barely conduct causal inference, especially counterfactual reasoning, leading to a failure of in-depth understanding of the acoustic mechanism behind.

In order to solve the problems of uneven clustering and poor energy consumption control of layered routing protocols in wireless sensor networks, an inter-cluster multi-hop hierarchical routing algorithm based on ISODATA clustering (IICMH) is proposed. Firstly, in the stage of network clustering, the ISODATA clustering algorithm is used to select the clustering center, and then the cluster head nodes are selected according to the central position of each cluster, and the election results are broadcasted by the base station. In terms of cluster head election mechanism, the algorithm sets an energy threshold T for comparing with cluster head to evaluate the specific situation of residual energy. When the value of residual energy is higher than that threshold, no new election is needed. On the contrary, it will trigger the next round of new elections, selecting new cluster heads and then update them to the base station in real time. The simulation results show that the algorithm has good performance in balancing network load and prolonging network lifetime.

Degradation kinetics of pharmaceutical excipient films effects the overall performance of drug releasing. However, characterizing the kinetics is traditionally labor-intensive and time-consuming, requiring spectroscopy or periodic mass measurements. Here we present an efficient alternative technique based on recent advances of deep learning and computer vision, in an effort to accurately measure the polymeric matrix swelling and material degradation in aqueous media in a timely manner. We incorporate a deep convolutional neural network (CNN) trained by a large-scale image dataset we collect, where a film is loaded with ferromagnetic nanoparticles and used as the core of a planar coil whose resonant frequency is monitored remotely. When placed in an aqueous solution, swelling and dissolution of the film induce contrasting changes in the capacitance and inductance of the coil, respectively, allowing identification of the swelling and dissolution phases. The dissolution profile of iron oxide-loaded gelatin is compared with spectrophotometry data, further demonstrating the technique can distinguish among films with various levels of crosslinking (showing a resonant frequency difference of 116 kHz between completely non-crosslinked and fully crosslinked gelatin). The key characteristics of the film degradation kinetics can be captured within 30 minutes of data collection.

The deformation of a horseshoe bat’s parietal plays an important role in its biosonar system, by understanding which we may gain an insight on the functionalities and geometric structures of the complex bionar system in the hope of building an artificial alternative with comparable features in the future. However, this promising research direction is not well investigated yet, limited by either the availabilities of experimental horseshoe bats or the capabilities of physical devices, namely, existing high-speed cameras fail to record videos of a quality enough for correlating parietal deformations and ultrasonic pulses in a fine-grain manner.

Colorectal cancer (CRC) is one of the malignant tumors with high morbidity and mortality. A prevalent method for studying colorectal cancer is to identify differentially expressed genes (DEGs) between control and patient samples, followed by the pathway enrichment analyses. However, many of those studies ignore the fact that different pathological stages of the cancer are often highly different from each other. The mixture of those heterogeneous samples may lack the efficiency of identifying the real DEGs, and loss the opportunity to analyze the dynamic evolution process of cancer. In this study, we develop a feasible framework to identify function evolution processes of cancers according to their pathological stages. Firstly, the limma package was used to identify DEG sets between control and CRC stage I, II, III, and IV samples, separately. Secondly, a pathway interaction network was constructed by taking a comprehensive analysis of pathways at individual stages, and a functional module interaction network was also generated independently by clustering genes into modules in a PPI network. The relationship between pathways and modules in adjacent stages was analyzed for all stages of CRC. A total of 479, 313, 349, and 383 DEGs were identified and they were enriched in 17, 16, 20, and 24 pathways, respectively. A functional evolution network was constructed by using those modules, and two significant evolution processes a2-b1-c2-d1 (Mod1) and a1-b2-c1-d2 (Mod2) were identified which may play critical roles in the development of CRC. The framework proposed in this study can be used to explore molecular mechanisms and evolution processes of CRC.

To improve existing online portfolio selection strategies when the number of assets is large, we propose an online portfolio selection strategy based on the proximal gradient projection (PGP). This paper first applies the L_1 norm of portfolio weights to portfolio construction and gains the sparse rebalance rule. PGP strategy is backtested on the four datasets. The empirical results show that the PGP strategy outperforms almost all the latest portfolio selection strategies in most cases. In addition to its superior performance, the PGP algorithm needs to be linear with the product of the number of stocks and trading days. It runs fast in backtests and thus is very suitable for large scale or high frequency online trading applications. Moreover, the SGP strategy can sustain reasonable transaction costs.