Search

Since most existing routings in socially aware networking only perform well in a certain statistic of network performance (e.g., Epidemic has high delivery rate but high network overhead, Direct Delivery has low network overhead but poor performance in latency and delivery rate), a new routing scheme: Community Correlation Based Routing (CCBR) is proposed in this paper to balance the delivery rate and network overhead. In CCBR, message forwarding is divided into two phases: in-community and across-community. We first define three new indicators by analyzing the social attributes of nodes: the social relationship between nodes, the active of node and the community correlation between communities. Then we use the three new indicators to form two functions: in-community forwarding utility and across-community forwarding utility. When messages are in the stage of in-community forwarding, we select the node with higher in-community forwarding utility as the relay node to make the messages deliver to the destination node faster. And when messages are in the across-community forwarding stage, the relay node with higher across-community forwarding utility is selected to avoid the messages being confined to the local community and make sure that the messages can always be transmitted to the destination community. Extensive simulations show that the proposed CCBR routing scheme can effectively improve the message delivery rate and greatly reduce the network overhead. And it can achieve better performance compared with the existing routing schemes even under the limited cache.

Infected areas and damage levels due to crop pest and disease have been growing seriously according to the climate change. We aim to develop an automatic system to provide national pest and disease dynamic monitoring and early forecasting products, by integrating multisource information (Earth Observation, meteorological, ecological, entomological, and plant pathological, etc.) and cutting edge research on pest and disease modeling to support decision making in the sustainable management of pest and disease. First, we selected the sensitive indexes for pest and disease habitat monitoring and early forecasting, and then optimized the forecasting model's parameters to enhance its applicability in national level. Second, we developed an automatic system based on web GIS platform to efficiently realize the national pest and disease dynamic habitat monitoring and early forecasting. Finally, we released the pest and disease forecasting thematic maps. China's national disease wheat yellow rust ( Puccinia striiformis ) and national pest oriental migratory locust ( Locusta migratoria manilensis (Meyen) ) are taking as the experimental objects. Based on the developed system, we forecasted the infected areas of rust and locust in China, in 2019, with these R-square values are higher than 0.87. This system would not only promote the efficacy of pest and disease management and prevention by improving accuracy of monitoring and forecasting, but also help to reduce the amount of chemical pesticides, which could thus guarantee food security and agriculture sustainable development in China.

We develop the method of machine learning for sub-pixel depth perception based on flexible and separated structured light, and describe novel framework for getting the extrinsic parameters of the system without complicated calibration in advance. We combined the extrinsic parameter calculation process into the structured light decoding process. PSO-C4.5 Decesion Tree algorithm is used to classify and match sub-pixel based on the Label. According to the matching result, the disparity map at the sub-pixel level is obtained, and the extrinsic parameters of the measurement system are obtained according to the fundamental matrix and initial pose information. In contrast to previous work, we formulate the problem of correspondence matching as a classification problem, which allows us to correctly take into account effects caused by scene movement. All optimization problems are solved with machine learning techniques. We test our method on a number of real-world examples as well as our new benchmark dataset for depth map estimation, and compare results to the state-of-the-art methods. The proposed method is both faster as well as more accurate.

In this paper, we mainly improved the grasp detection network based on the grasp pose detection (GPD) algorithm. Three Network in Network (NIN) structure blocks are used as feature extraction modules, and a fully connected layer is used for classification. And trained the network on our own dataset which these 3D models are all from GraspNetAPI, and the accuracy rate on the test set is 2% higher than the original network. And the amount of its parameters has dropped by 38%. In addition, we combine the Mask R-CNN semantic segmentation network to perform target detection and pixel-level instance segmentation on target objects, which can grasp objects of interest. Finally, we carried out grasping experiments on real robots, and the total grasping success rate was over 70%.

The control performance of converter directly affects the drive efficiency of electric vehicle (EV). This paper studies on the model predictive control (MPC) based controller design of battery integrated modular multilevel converter (B-MMC). Aiming at the goal of three-phase current tracking and battery balancing control of B-MMC, three-layer controllers are designed. The first controller is the finite control set model predictive control (FCS-MPC), which realizes the motor drive by tracking the phase current. The second controller is the circulation controller for the active balance control of battery. It is composed of multiple proportional controllers to realize the battery balance between the bridge arms. The third controller is the battery sorting algorithm to achieve the passive balance control of the battery in the same bridge arm. At the same time, field programmable gate array (FPGA) is used to accelerate the controller to meet the real-time requirements of B-MMC. Finally, the simulation results of the B-MMC control system validates the effectiveness of the control algorithm.

Based on the predominance of Multi-Agents Simulation in evacuation simulation, the simulation algorithms Based on RL and SFM (Social Force Model) is designed in this paper to accommodate emergency evacuations. The A*, A* with SFM are compared with algorithms we designed. The result presented analytically and graphically, results indicate A* algorithm has theoretically optimal paths, but algorithms Based on RL and SFM is closer to human behavior patterns.

The task of matching between images acquired by terminal equipment and satellites is important and challenging due to the dramatic viewpoint changes and unknown orientations, especially with different imaging sensors. In this paper, we firstly present the task to match the optical/infrared images acquired by UAVs with satellite SAR images. Many previous works mainly focused on matching with the images of the same modality, and may not perform well to our task. To overcome the difficulties caused by the diversity among the three modalities of data, we mine the common features of them and propose a novel edge-driven matching framework to find the correspondence between the UAV images and SAR images. Experimental results demonstrate the effectiveness and superiority of our method.

Energy management strategy is the key problem of hybrid electric vehicles (HEV). In this paper, a hierarchical control of energy management strategy of HEV is proposed. In the higher layer, which is named as velocity trajectory optimization layer, the optimal velocity is obtained by considering the slope of the road ahead. In the lower layer called the torque split layer, the power split and gear ratio based on model predictive control (MPC) are optimized by applying the velocity obtained from the higher layer. Simulation results show that the proposed hierarchical control strategy can effectively improve the fuel economy of the HEV.

In this paper, a support vector machine (SVM) based model predictive control (MPC) strategy is proposed to enable the vehicle to track the reference path. Considering that learning nonlinear process using SVM is simple and the results are accurate and without over-fitting,the predictive model is trained by SVM. In order to retain the dynamic information on the basis of reducing the computational burden, the instantaneous linearization is applied to SVM model by Taylor expansion. Moreover, its accuracy is verified. Then the model predictive controller is obtained to achieve the vehicle path tracking control. Finally, the simulation results demonstrates the control performance of the proposed method.

For the problem of vehicle obstacle avoidance control, a steering obstacle avoidance method based on Gauss pseudospectral method is designed. In this paper, the kinematics and dynamics models of vehicles are established as dynamic constraints, combining with state constraints and path constraint, the obstacle avoidance behavior of intelligent vehicles is summarized as a continuous optimal control problem which takes obstacle avoidance time as the optimal objective. In the research of solving the optimal control problem, Gauss pseudospectral method is adopted which utilizes numerical approximation techniques to transform the continuous optimal control problem into a discrete nonlinear programming problem with series of equality and inequality constraints. The trajectories of control variable and state variable are obtained through solving this nonlinear programming problems numerically. The effectiveness and control performance of this method are illustrated by simulation.

An ultra wideband Vivaldi antenna array based on tightly coupled effect is proposed. The elements are arranged overlap to get an additional capacitive capacity to achieve impedance matching at low frequency. A sample of a 4× 4 Vivaldi array was simulated, fabricated, and tested. The measured farfield waveform is basically agreed with simulated one. The designed antenna gets a bandwidth of O.4GHz-4.2GHz.

Fire prevention and control had significant meaning for public safety and social development. To realize automatic monitoring of compartment fire, this paper proposed an intelligent indoor fire detection method based on infrared thermal image. The first step in the process was to locate and detect suspicious areas in the infrared image. Then the Principal Component Analysis method was utilized to extract features and reduce the dimension of feature. Finally, a Support Vector Machine classifier was designed and trained to distinguish a potential flame from a fire and a light. Compared with k-nearest neighbor (KNN) classifier, Random Forest(RF) classifier, and Logical Regression(LR) classifier, SVM classifier had better performance. The accuracy rate of SVM classifier in the test set was 99.97%, and the flame recall rate by SVM was 99.996%. Experimental results demonstrated that the flame detection method proposed in this paper had significant detection effect and good application prospects.

In order to establish the kinematics and dynamics modeling of the quadruped robot, a dynamics modeling approach based on screw theory is demonstrated. Kinematics and the Jacobian matrix of single chain are obtained at first. Then, the dynamic modeling of single chain was established. All the 4 single chains are clustering together, and we get dynamics equation of the quadruped robot. The quadratic program solve an optimal control problem to track linear and angular acceleration of the body as closely as possible under the premise of satisfying a series of constraints of the full quadruped robot dynamics. Lastly, simulations and physical prototype experiment have been done to verify the effectiveness of the proposed method.

With the development of cloud computing, the reliability of disk arrays are increasingly concerned. Data centers usually use erasure codes to provide high reliability. However, most of reconstruction methods on disk arrays focus on single/ multiple disk(s) recovery, which ignores how to efficiently reconstruct the lost data such as Latent Sector Errors (LSEs), etc. In real situations, local stripe errors are much more common than disk failures. It has become an urgent problem that how to improve reconstruction efficiently for stripes. This paper proposes a comprehensive rearranging priority reconstruction(CRPR), which combines temporal locality, spatial locality and coding characteristics together. CRPR divides different blocks into various priorities and recovers them sequentially. To demonstrate the effectiveness of CRPR, we conduct several simulations via disksim. The simulations results show that, the comprehensive rearranging priority reconstruction method keeps up with previous methods and can save up to 63.9% in terms of waiting time.

Human-Object Interaction (HOI) Detection is an important problem which is expected not only to detect individual object instances, but also to recognize the visual relationship between object pairs. In this paper, we improve the performance of the HOI detection task based on the instance-centric attention network. At first, we improve the accuracy of existing model by optimizing loss function and training details. We validate the improved method on the recently introduced Verbs in COCO (V-COCO) dataset. Then, to analyze the experimental results, we set several error modes and get the distribution of the false positives for each action class. Finally, we apply the trained model to HOI detection in surveillance scenario. The experimental result shows that the improved method can accurately detect the human-object interactions, which are involved in the VCOCO dataset.

In the present work, this paper demonstrated that pore is an alternative additive for obtaining a substrate material with equivalent permeability and permittivity by decreasing permeability of NiCuZn ferrite ceramic. Firstly, the (Ni 0.2 Cu 0.2 Zn 0.6 O) 1.03 (Fe 2 O 3 ) 0.97 ferrite ceramic with 0.25 wt.% Nb 2 O 5 and 0.50 wt.% Bi 2 O 3 additives was synthesized by a solid phase method at O 2 atmosphere. Results showed that the compact NiCuZn ferrite (c-NiCuZn) sintered at 900 °C possess a high permeability (μ r ’=427.6) and low magnetic loss (tanδ μ =0.016) at 1MHz. Also, the SEM image showed that the ferrite possess uniform and compact microstructures. As a reference, the NiCuZn ferrite without any additives (named n-NiCuZn) was obtained by same method. Results indicated the permeability of the n-NiCuZn ferrite ceramic with few pores obviously reduced (μ r ’=151.4) at 1MHz due to low sintering temperature and inadequate reaction. Meanwhile, the magnetic loss (tanδ μ ) is 0.029 at 1MHz. To obtain porous NiCuZn (p-NiCuZn) ferrite ceramic, we added 10.0 wt.% g-C 3 N 4 into the pure NiCuZn ferrites. Due to remove of g-C 3 N 4 powders (high temperature sintering), there are more pore between grains of the NiCuZn ferrite. Thus, the permeability and magnetic loss of the p-NiCuZn are 53.7 and 0.041, respectively. More importantly, the cutoff of the p-NiCuZn ferrite ceramic can increase from 10.8 MHz to 63.1 MHz when more pores were introduced. Thus, the pores are an alternative additive to reduce permeability of magnetic ferrite ceramics to meet application requirements.

Infected areas and damage levels due to pest and disease have been growing seriously according to the global climate changes. The government department of plant protection normally collects the crop pest and disease information by manual inspection, which is unable to provide timely and spatially continuous crop growth status and pest/ disease development over large areas. The manuscript aims to bring together and produce cutting edge research to provide crop pest and disease monitoring and forecasting information, integrating multi-source (Earth Observation-EO, meteorological, entomological and plant pathological, etc.) to support decision making in sustainable management of pest and disease. Taking national disease –a fungal disease of wheat rust and national pest – a serious insect pest locust as the experimental object, we conducted the following research: 1) the sensitive spectral features of rust and locust with capability of stresses differentiation would be identified or formed based on field hyperspectral data and UAV hyperspectral images for detection, 2) multi-sources of data that are composed by remote sensing images (eg. GF, Landsat and Sentinel) and meteorological observations would be integrated to evaluating habitat of rust and locust in farmland level, 3) multi-temporal remote sensing images and vegetation ecological dataset are combined to provide environmental information of rust and locust dispersion, and forecast the damaged areas and levels in regional scales. Moreover, an automatic system is developed to do the disease and pest timeseries monitoring and forecasting, also the visual display of the thematic maps and analysis reports. The system is constructed based on WebGIS platform. It selected Browser/Server (B/S) architecture, access the system interface through a web browser, simple, quick and easy to operate. And, it could timely, efficiently, quantitatively do the extraction and analysis of crop pest and disease occurrence and developing information, also produce pest and disease thematic maps and scientific report. The crop pest and disease monitoring and forecasting system, can provide effective information of pest and disease developing for our agricultural sector, provide a scientific basis to formulate pest and disease prevention and control measures, and also provide data basis and technical support for the crop network management. Based on the system, we analysed the damaged situation and changes of national wheat rust in 2019, and locust in Tianjing 2019. The results would not only promote efficacy of pest and disease management and prevention by improving the accuracy of monitoring and forecasting, but also help to reduce the amount of chemical pesticides, which could thus guarantee the food security and sustainable development of agriculture in China.

Searching targets of interest in large-scale satellite images is an imperative task, which becomes a challenging issue when the targets reside only in the mind of the user as a set of subjective visual patterns. In this paper, we take the advantage of hand-drawn sketches’ strong intuition of describing mental target to address the problem of no available exemplar query. We introduce a multi-level-of-detail model to learn a cross-domain representation for bridging the gap between sketches and satellite images. To train the model, we propose a novel method of generating satellite images with corresponding level of details based on generative adversarial network. Experiments on both large-scale satellite images and commonly used RS datasets demonstrate the effectiveness and superiority of our method.

Power line detection plays an important role in automated UAV inspection system, which is crucial for real-time motion planning and navigation along power lines. Previous methods which adopt traditional filters and gradients may fail to capture complete power lines due to noisy background. To overcome this, we develop an accurate power line detection method using rich convolutional and structured features. The proposed method fully exploits multiscale and structured prior information to conduct both accurate and efficient detection. We evaluate the method on two well-annotated power line datasets and achieve state-of-the-art performance compared with previous methods.