Modern intelligent ships are evolving into highly collaborative, complex, and tightly integrated systems. Diverse operation modes emerge as different mission requirements, which lead to increased demands for precision and reliability in control. In this paper, a Long Short-Term Memory (LSTM) assisted Model Predictive Control (MPC) method is proposed to fit each operating condition of trajectory tracking for Unmanned Surface Vehicles (USVs). Firstly, the LSTM network is utilized to predict the behaviors of the controlled system, inspired by the robust representational capabilities of deep learning. Based on this mechanism, the different operation modes of USVs could be automatically matched without the switching strategy or building multiple models for each mode. Then, additional collision avoidance constraints are added to the MPC to enhance safety measures and address the optimization problem. Finally, a case study including three different trajectory tracking scenarios is conducted to validate the proposed method. The simulation results show that the model is proven to have the high control accuracy and effective collision avoidance response within the confines of obstacles under the disturbed environment. The model could be used to provide invaluable guidance to decision making such as the remote control of human-computer interaction and maritime rescue.

Circular braiding is a composite preform manufacturing technology. The structural parameters such as braid angle of composite preform have great influence on its mechanical properties. In the circular braiding system, a robot move the mandrel with a certain trajectory and speed to control the braid angle. Aiming at the problems of unstable braid angle in the process of braiding complex mandrel, a trajectory planning method of robot is proposed. Firstly, a yarn deformation model in process of circular braiding is developed to calculate the real-time actual convergence zone length. Then, based on the variation of the convergence zone length and the spatial geometric relationship that the center line of the mandrel is always perpendicular to the braiding plane, the trajectory points of robot in the mandrel coordinate system are solved. Finally, the trajectory points in the robot coordinate system are obtained by a series of coordinate system transformations. The experimental results show that, with corresponding take-up speed profile, the trajectory planning method can effectively reduce the braid angle error of curved mandrel and variable cross-section mandrel within ± 2°, which can improve the quality and efficiency of composite preform production.

This paper addresses the problem of distributed energy management in multi-area integrated energy systems (MA-IES) using a multi-agent deep reinforcement learning approach. The MA-IES consists of interconnected electric and thermal networks, incorporating renewable energy sources and heat conversion systems. The objective is to optimize the operation of the system while minimizing operational costs and maximizing renewable energy utilization. We propose a distributed energy management strategy that makes hierarchical decisions on intra-area heat energy and inter-area electric energy. The strategy is based on a multi-agent deep reinforcement learning framework, where each agent represents a component or unit in the MA-IES. We formulate the problem as a Markov Decision Process and employ Q-learning with experience replay and double networks to train the agents. The proposed strategy is evaluated using a simulation of a four-area MA-IES. The results demonstrate significant improvements in energy management compared to traditional methods, with higher renewable energy utilization and lower operational costs. Specifically, the strategy achieves 100% utilization of wind power, and decreases operational costs by 5.53%. Furthermore, it leverages the generalization capabilities of reinforcement learning to respond in real-time to uncertainties in demand and wind power output. The results highlight the advantages of the proposed strategy, making it a promising solution for optimizing the operation of multi-area integrated energy systems.

Circular braiding enables efficient production of fiber reinforcements. The mechanical properties of fiber reinforcements are largely affected by the braiding angle, so controlling the braiding angle by adjusting the mandrel traction speed is a key issue in circular braiding. This paper proposes an algorithm for generating and correcting the mandrel traction speed. First, the mandrel model and the dynamic braiding model are established, the state changes in convergence zone are analyzed and the initial traction speed is calculated. Then, the traction speed is corrected during braiding based on the braiding angle detection. To handle detection delay, the braiding process is predicted based on the known mandrel model. Besides, the braiding model parameters are updated according to detection results to improve modeling accuracy. Simulation tests show that using this method can reduce braiding angle error compared to the baseline. In the case without external interference, the maximum error is reduced by 10.3%, and in the case with external interference, the maximum error is reduced by 7.8%.

Overbraiding allows rapid production of complex composite performs, especially using the carbon fiber. There are a variety of geometric and mechanical models to describe the overbraiding process. However, the existing models are either low accuracy, limited to axisymmetric mandrels or high time cost, which means the existing methods are not applicable for closed-loop control during overbraiding. This paper proposes a new model to describe the overbraiding process of non-axisymmetric mandrels with low time cost and high accuracy, and this model is applicable for closed-loop control. The model is based on the enhanced kinematic models and uses the Newton-Raphson method to solve the equations, which ensures the low time cost. Besides, the relationship between the interlacement of yarns in the convergence zone and the deposition of yarns on the mandrel are considered as a whole in this paper, which improves the accuracy.

Passive detection can work for a long time with low energy consumption in underwater surveillance. However, tracking unknown noncooperative targets with only direction angles is challenging, and the tracking performance of multiple targets is poor. Based on several passive sensors in the underwater sensor network (UWSN), a feature-aided state estimation method is used to start tracking unknown targets. The feature-aided joint probabilistic data association combined with the particle filter method is also proposed to improve the passive tracking performance of multiple targets. The track management and the fusion strategy are given to remove fake tracks and obtain correct trajectories of unknown targets. The simulation results show that the feature-aided method can quickly start and effectively track multiple noncooperative targets with passive sensors. Compared with other methods, the proposed method can track targets more accurately with the advantages of low energy consumption and less exposure in various environments.

Robots need more intelligence to complete cognitive tasks in home environments. In this paper, we present a new cloud-assisted cognition adaptation mechanism for home service robots, which learns new knowledge from other robots. In this mechanism, a change detection approach is implemented in the robot to detect changes in the user’s home environment and trigger the adaptation procedure that adapts the robot’s local customized model to the environmental changes, while the adaptation is achieved by transferring knowledge from the global cloud model to the local model through model fusion. First, three different model fusion methods are proposed to carry out the adaptation procedure, and two key factors of the fusion methods are emphasized. Second, the most suitable model fusion method and its settings for the cloud-robot knowledge transfer are determined. Third, we carry out a case study of learning in a changing home environment, and the experimental results verify the efficiency and effectiveness of our solutions. The experimental results lead us to propose an empirical guideline of model fusion for the cloud-robot knowledge transfer.

海洋船舶目标识别在民用和军事领域有着重要的战略意义,本文针对可见光图像和红外图像提出了一种基于注意力机制的双流对称特征融合网络模型,以提升复杂感知环境下船舶目标综合识别性能.该模型利用双流对称网络并行提取可见光和红外图像特征,通过构建基于级联平均融合的多级融合层,有效地利用可见光和红外两种模态的互补信息获取更加全面的船舶特征描述.同时将空间注意力机制引入特征融合模块,增强融合特征图中关键区域的响应,进一步提升模型整体识别性能.在VAIS实际数据集上进行系列实验证明了该模型的有效性,其识别精确度能达到87.24％,综合性能显著优于现有方法.

充分考虑水下环境和水下成像的特点,将多尺度融合的图像增强算法应用于水下配准算法的预处理图像中,修复深水域偏色严重的图像.用改进SIFT算法进行特征提取,采用自适应阈值法筛选关键点,扩大关键点提取范围;用Canny算法计算关键点的梯度和大小,平滑噪声的同时也可以保留图像更多细节;使用平均Haus-dorff距离和BBF最邻近查询法对关键点进行粗匹配,再用RANSAC进行进一步提纯;计算得到变换矩阵后输出最终拼接图像.实验验证了该算法适合水下环境特点,可以提升水下图像配准和拼接的效果和准确率.

In order to enable robots to provide high-level services in complex indoor environments, it is necessary to improve the robots' ability to cognize the environments. Most of the existing research is focused on indoor 3D reconstruction and semantic segmentation without the organization and maintenance of object recognition results. In this paper, we present an approach to build a 3D semantic map that includes both voxel-based geometrical demonstrations and object-aware entities with the combination of Simultaneous Localization and Mapping (SLAM) and Mask Region-based Convolutional Network (Mask R-CNN). An extended seeded region growing algorithm is designed for 3D segmentation refinement, and an octree-based framework octomap is used to present 3D map in replacement of point cloud map. We present experiments in a simulated home environment and the experimental results verify the accuracy and efficiency of our method.

Maneuvering target tracking is one of the most important topics in marine research, and it is vital to track a target accurately. Tracking accuracy is closely related to the veracity of estimation of target's motion model. To improve the accuracy of tracking single underwater maneuvering target, this paper brings up a new method to reduce tracking error caused by switching of motion models. A support vector machine (SVM) is used to classify current motion model of the target, then state of the target is estimated by a Kalman filter (KF) with dynamic parameters. Simulation results suggest that compared with the classical interacting multiple model (IMM), algorithm proposed in this paper leads to a more satisfactory tracking root mean square error (RMSE) no matter the target maneuvers or not.

Smart homes can provide complementary information to assist home service robots. We present a robotic misplaced item finding (MIF) system, which uses human historical trajectory data obtained in a smart home environment. First, a multi-sensor fusion method is developed to localize and track a resident. Second, a path-planning method is developed to generate the robot movement plan, which considers the knowledge of the human historical trajectory. Third, a real-time object detector based on a convolutional neural network is applied to detect the misplaced item. We present MIF experiments in a smart home testbed and the experimental results verify the accuracy and efficiency of our solution.

为了提高提花织物纹样提取的准确性,消除织物组织结构对提取结果的干扰,文章提出了一种基于纹理消除滤波算法和密度峰聚类算法的纹样提取方法.通过统计图片区域梯度信息分离提花织物组织纹理区域和图案边缘结构,并利用非极大值抑制获得纹样边缘结构作为滚动引导滤波器的引导图,经过多次迭代计算实现织物纹理滤波.然后将图片从RGB颜色空间转换为CIELab颜色空间,利用密度峰聚类算法对织物色彩空间分割聚类,最终提取出织物纹样.实验结果表明,文章提出的方法快速准确地实现了提花织物纹样的自动分割与提取.

In order to solve the multiple waypoints path planning problem in smart home environment, we adapt the optimal sampling-based algorithm (RRT*) [1] to deal with multiple waypoints navigation, namely Multi-RRT*. Our method constructs multiple trees from multiple waypoints, and these trees use simple extension and connection strategy. When all trees are merged to a single tree, an traversal path will be found. Along this path, mobile robot can visit all the waypoints one by one. We evaluate this method on a designed path planning benchmark scenario and compare with the basic and bias RRT*. Finally, we apply the proposed algorithm on Turtlebot2 with Robot Operate System (ROS) in smart home experimental environment. Simulation and experimental results demonstrate that our algorithm can converge quickly and has good performance in dealing with multiple waypoints path planning problem.

With the high development of optical fiber communication and manufacturing of related high-precision products,traditional detection methods based on hands,eyes and so on cannot only cause great error,but also do not meet the demand of large-scale production.Through improving the traditional mathematical model based on gray-value,and building shape-based mathematical model suitable for real target detection,we presented the overall detection algorithm of one fiber transceiver board PCB.The result shows the measurement error of the algorithm in 300 sample products can be controlled within 5 microns,and the average measuring time of every product can be kept within 200 microseconds,which meets the needs of mass production.The algorithm has better recognition performance comparing to traditional pattern matching and it has important significance to the design and development of automatic detection system of similar high-precision products.

This paper addresses the latency of visual feedback in an immersive telepresence robotic system. The latency is defined as the delay between the real event occurring in the robot's environment and the event displayed to the user. While it has been a widely recognized problem in teleoperation and telepresence, the existing research mainly focuses on mitigating it and there are relatively few works that conducted quantitative measurements. As opposed to most existing works, we propose to quantify and compare the latency caused by each internal connection in telepresence robotic systems. Our proposed method is able to identify key contributors to the overall latency, and the experimental results can be used to design effective mitigating strategies.

Underwater wireless sensor networks (UWSNs) can provide a promising solution to underwater target tracking. Due to limited energy and bandwidth resources, only a small number of nodes are selected to track a target at each interval. Because all measurements are fused together to provide information in a fusion center, fusion weights of all selected nodes may affect the performance of target tracking. As far as we know, almost all existing tracking schemes neglect this problem. We study a weighted fusion scheme for target tracking in UWSNs. First, because the mutual information (MI) between a node’s measurement and the target state can quantify target information provided by the node, it is calculated to determine proper fusion weights. Second, we design a novel multi-sensor weighted particle filter (MSWPF) using fusion weights determined by MI. Third, we present a local node selection scheme based on posterior Cramer-Rao lower bound (PCRLB) to improve tracking efficiency. Finally, simulation results are presented to verify the performance improvement of our scheme with proper fusion weights.

Silk relics have a long history, and its color matching contains great cultural value which is of great significance for the research and utilization of the color information of the silk relics.However,it is time-consuming and laborious to use the traditional manual method to achieve the color matching process.Aiming at this problem,the color matching expert system based on silk relics is proposed.The system uses the information of silk relics(including age information,subject information,tone information)and domain expert knowledge to design different color schemes.User data is acquired by the front end user interface,then the data is transferred to the back-end reasoning module,and the reasoning module is implemented by Prolog language, and expert knowledge is described by traditional fuzzy rules, finally,the matching color scheme is provided to the user.The results show that the system greatly improves the utilization efficiency of silk relics and the efficiency of silk designer.

针对现有机织物组织识别方法适用范围窄、鲁棒性差的现状,课题组提出一种计算织物组织循环大小的平移相减算法(translational subtraction algorithm,TSA),并提出了一种基于TSA算法的机织物组织有效识别方法.该方法结合机织物图像不同方向的TSA算法和水平方向的亮度累加法获取织物组织循环宽度和纬线宽度,然后对机织物图像进行错位TSA算法,分析错位TSA曲线相位的周期性和大小,可以得到织物组织循环的纱线根数和飞数,最终获得机织物图像的组织意匠图.实验证明该方法对机织物图像光照、纹理和倾斜等干扰因素具有鲁棒性,能有效识别各种类型的机织物组织.