检索结果-内蒙古大学图书馆

EK-Net++: Real-time scene text detection with expand kernel distance and Epoch Adaptive Weight

Expert Systems with Applications 2025年 267卷

作者： Zhu, Boyuan Chen, Xi Tang, Quan Chen, C.L. Philip Liu, Fagui Computer Science and Engineering South China University of Technology Guangzhou510006 China Cyberspace Security Research Center Peng Cheng Laboratory Shenzhen518066 China State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100080 China

Recently, scene text detection has received significant attention due to its wide applications. Accurate detection in complex scenes of multiple scales, orientations, and curvature remains a challenge. Component-based methods have emerged as the dominant approach for offering a more flexible way to achieve superior performance. However, Component-based methods employ the Vatti clipping algorithm for multi-instance training to handle the difficulty of arbitrary-shaped text. Yet we identify several bias results from these approaches called the "shrinked kernel" leading to decreased accuracy. Additionally, almost all text detection methods fail to consider the characteristics of different datasets and directly assign fixed loss weights to multi-instances. To address these limitations, we propose a new approach named Expand Kernel Network (EK-Net++), which presents the expand kernel distance map to compensate for the previous deficiency and incorporates three-stage regression to complete integral instance detection. Furthermore, we design the Epoch Adaptive Weight algorithm which dynamically distinguishes the characteristics of different datasets. EK-Net++ not only achieves precise positioning of arbitrary-shaped text but also strikes a balance between performance and speed. Evaluation results show that EK-Net++ attains state-of-the-art or competitive performance compared to other advanced methods, e.g., F-measure of 86.72% at 35.42 FPS on ICDAR 2015, F-measure of 85.83% at 40.13 FPS on CTW1500. © 2024 Elsevier Ltd

关键词： Epoch Adaptive Weight Expand kernel distance Real-time Scene text detection

来源：评论

学校读者我要写书评

暂无评论

RoboDesign1M: A Large-scale Dataset for Robot Design Understanding

arXiv

引用

arXiv 2025年

作者： Le, Tri Nguyen, Toan Tran, Quang Nguyen, Quang Huang, Baoru Nguyen, Hoan Vu, Minh Nhat Ta, Tung D. Nguyen, Anh FPT Software AI Center Viet Nam Department of Computer Science University of Liverpool United Kingdom University of Information Technology HCMC Viet Nam Automation & Control Institute TU Wien Austria Department of Creative Informatics The University of Tokyo Japan

Robot design is a complex and time-consuming process that requires specialized expertise. Gaining a deeper understanding of robot design data can enable various applications, including automated design generation, retrieving example designs from text, and developing AI-powered design assistants. While recent advancements in foundation models present promising approaches to addressing these challenges, progress in this field is hindered by the lack of large-scale design datasets. In this paper, we introduce RoboDesign1M, a large-scale dataset comprising 1 million samples. Our dataset features multimodal data collected from scientific literature, covering various robotics domains. We propose a semi-automated data collection pipeline, enabling efficient and diverse data acquisition. To assess the effectiveness of RoboDesign1M, we conduct extensive experiments across multiple tasks, including design image generation, visual question answering about designs, and design image retrieval. The results demonstrate that our dataset serves as a challenging new benchmark for design understanding tasks and has the potential to advance research in this field. RoboDesign1M will be released to support further developments in AI-driven robotic design automation. © 2025, CC BY-NC-ND.

关键词： Machine design

来源：评论

学校读者我要写书评

暂无评论

Node-Decomposition Privacy-Preserving for Finite-Time Secondary control of DC Microgrids

引用

IEEE Transactions on Smart Grid 2025年

作者： Wu, Han Zhu, Si-Min Chai, Li Zhu, Zhen-Hua Tian, Yu-Chu Wuhan University of Science and Technology Engineering Research Center of Metallurgical Automation and Measurement Technology School of Artificial Intelligence and Automation Wuhan430081 China Zhejiang University State Key Laboratory of Industrial Control Technology College of Control Science and Engineering Hangzhou310027 China Huzhou Institute of Industrial Control Technology Huzhou China Queensland University of Technology School of Computer Science BrisbaneQLD4001 Australia

In the distributed secondary control of DC microgrids, achieving both rapid convergence and robust information security is critical for enhancing control performance and maintaining power quality. However, current privacy-preserving approaches often compromise either control performance or privacy preservation performance. To address this challenge, this paper presents a node-decomposition privacy-preserving (NDPP) method for fully distributed, finite-time secondary control in DC microgrids. By splitting each node in the communication graph into multiple sub-nodes, the method enhances both convergence speed and privacy preservation without sacrificing control accuracy. Sub-nodes communicate within groups, and the topology of the generated graph is concealed from malicious nodes and eavesdroppers. Theoretical analysis is provided to prove finite-time voltage regulation and current sharing with privacy preservation against malicious nodes and external eavesdroppers. Simulations are conducted to verify the effectiveness of the method in maintaining control performance and power quality. © 2010-2012 IEEE.

关键词： Privacy by design

来源：评论

学校读者我要写书评

暂无评论

Data-Driven Learning and control with Event-Triggered Measurements

引用

IEEE Transactions on Automatic control 2025年

作者： Feng, Shilun Shi, Dawei Chen, Tongwen Shi, Ling Beijing Institute of Technology State Key Laboratory of Intelligent Control and Decision of Complex Systems MIIT Key Laboratory of Servo Motion System Drive and Control School of Automation Beijing100081 China University of Alberta Department of Electrical and Computer Engineering EdmontonABT6G 1H9 Canada Hong Kong University of Science and Technology Department of Electronic and Computer Engineering Clear Water Bay Kowloon Hong Kong

Event-triggered control has attracted considerable attention for its effectiveness in resource-restricted applications. To make event-triggered control as an end-to-end solution, a key issue is how to effectively learn unknown system dynamics from event-triggered measurements and consequently, develop a learning-based event-triggered controller. Existing works learn system dynamics based on periodic time-triggered measurements, and it is yet to know how to learn a controller with performance guarantee based on event-triggered measurements. To address this issue, we consider the problem of learning an event-triggered state feedback controller for an unknown linear system based on event-triggered state measurements in this work. In particular, we first analyze the event-triggered measurements within a set-membership framework. We prove that the estimation error belongs to a bounded ellipsoid determined by the historical measurements and the event-triggering condition. Subsequently, we demonstrate that all admissible systems compatible with the collected data samples can be explicitly represented in the form of quadratic matrix inequalities using the state estimates. With the acquired set of admissible systems, a co-design problem for the data-driven controller and event-triggering condition is solved using the linear matrix inequality technique, with guaranteed closed-loop stability and L2-gain performance. Finally, numerical examples and comparisons are provided to illustrate the effectiveness of the proposed event-triggered learning and control approach. © 1963-2012 IEEE.

关键词： Networked control systems

来源：评论

学校读者我要写书评

暂无评论

A novel internal crack detection method for photovoltaic (PV) panels based on the air-coupled ultrasonic

引用

Sensors and Actuators A: Physical 2025年 392卷

作者： Sun, Tianyi Bai, Tianyang Cao, Shengxian Fan, Siyuan Zhang, Yanhui School of Automation Engineering Northeast Electric Power University Jilin132012 China Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen518055 China School of Control and Computer Engineering North China Electric Power University Beijing102206 China

Accurately assessing the potential risk of cracks in photovoltaic (PV) panels is crucial for improving the system's energy conversion efficiency and safety. This paper develops a novel internal crack detection device for PV panels based on air-coupled ultrasonics and establishes a dedicated model for PV panel crack detection. Considering the anisotropy of silicon, the Rayleigh–Lamb equation is first used to obtain the dispersion curve of Lamb waves. The continuous wavelet transform (CWT) is then applied to analyze the A0 mode Lamb wave echo signals within a single cycle to distinguish between cracked and non-cracked regions. Further, the original signal is enhanced using singular value decomposition (SVD) to strengthen the signal features, and a direct relationship model between the signal echo and the crack length is proposed. Finally, the model's validity is verified by measuring actual crack data from PV panels. The results show that the optimal frequency of Lamb waves for crack detection in PV panels is 200 kHz, with an attenuation coefficient of 6.4 dB/mm and incidence and exit angles of 11°. Compared to electroluminescence (EL) and inductive thermal imaging methods, the proposed approach achieves higher accuracy, reaching 84.26%. © 2025

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

Torque Characteristic Analyses and Multi-Objective Optimisation of Multi-Layer Flux-Barrier Less-Rare-Earth Permanent Magnet Synchronous Machine

引用

IET Electric Power Applications 2025年第1期19卷

作者： Wang, Weinan Gong, Weize Zhu, Liangkuan Wei, Jian Wang, Mingqiao Liu, Yong Zheng, Ping College of Computer and Control Engineering Northeast Forestry University Harbin China College of Mechanical and Electrical Engineering Northeast Forestry University Harbin China School of Electrical Engineering and Automation Harbin Institute of Technology Harbin China

This paper investigates the torque characteristic and multi-objective optimisation (MOO) of the multi-layer flux-barrier less-rare-earth permanent magnet synchronous machine (MLFB-LRE-PMSM) used for the electric vehicles (EVs). This study explores the variation of torque characteristics with current angle under different winding current conditions, and thoroughly analyses the influence of permanent magnet (PM) structure parameters on reluctance torque, PM torque and the proportion of reluctance torque in the electromagnetic torque. On this basis, the sensitivity analysis method based on Sobol sequence and joint Sobol index is adopted, which not only simplifies the traditional analysis process, but also effectively considers the interaction effect between the optimisation objectives. Finally, a cooperative optimisation strategy of improved whale optimisation algorithm (WOA) and genetic algorithm (GA) is proposed, which is successfully applied to the MOO design of MLFB-LRE-PMSM. The results show that the improved WOA algorithm shows excellent optimisation performance and can be used as a new solution in the field of motor optimisation. At the same time, the engineering practicability of the proposed collaborative optimisation scheme is verified by finite element simulation. © 2025 The Author(s). IET Electric Power Applications published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology.

关键词： optimisation permanent magnet machines sensitivity analysis

来源：评论

学校读者我要写书评

暂无评论

A Switched Neural Process for Long-Term Postprandial Glucose Trajectory Prediction Based on Insulin Sensitivity Estimation

引用

IEEE Transactions on Instrumentation and Measurement 2025年 74卷

作者： Chen, Jing Shi, Dawei Cai, Deheng Liu, Wei Ji, Linong Shi, Ling Beijing Institute of Technology MIIT Key Laboratory of Servo Motion System Drive and Control School of Automation Beijing100081 China Peking University People’s Hospital Department of Endocrine and Metabolism Beijing100044 China Hong Kong University of Science and Technology Department of Electronic and Computer Engineering Kowloon Hong Kong

Advanced sensing technology has enabled the real-time monitoring of blood glucose (BG), making it possible to explore the BG prediction problem based on continuous observations. In this work, we propose a novel switched neural process (SNP) approach for long-term postprandial BG prediction, which has the probabilistic inference ability conditioned on preprandial BG measurements. Distinct representative neural process (NP) models are constructed to capture the nonlinear BG dynamics based on a designed indicator. This indicator is developed to estimate insulin sensitivity (IS) at meal time by using physiological models, in which the effect of preprandial BG variation on IS estimation is evaluated. Besides, a robust model switching mechanism is designed to determine the appropriate NP model for prediction. The effectiveness of the proposed method is illustrated through the comparative results by using both the in silico data from the U.S. Food and Drug Administration (FDA)-accepted UVA/Padova type 1 diabetes mellitus (T1DM) simulator and the clinical data from 47 T1DM subjects. The obtained results indicate that the proposed method outperforms existing methods, especially for 3- and 4-h prediction, which shows promising potential in predicting postprandial BG levels. © 1963-2012 IEEE.

关键词： Glucose

来源：评论

学校读者我要写书评

暂无评论

Deep Reinforcement Learning of Robotic Manipulation for Whip Targeting

Deep Reinforcement Learning of Robotic Manipulation for Whip...

引用

IEEE International Conference on Simulation, Modeling, and Programming for Autonomous Robots (SIMPAR)

作者： Xiang Bai Junyi Wang Xiaofeng Xiong Evangelos Boukas Applied Mathematics and Computer Science Technical University of Denmark Copenhagen Denmark Automation and Control Group Electrical and Photonics Engineering Technical University of Denmark Copenhagen Denmark SDU Biorobotics Mærsk Mc-Kinney Møller Institute University of Southern Denmark Odense Denmark

ISBN: (数字)9798331516857

ISBN: (纸本)9798331516864

This paper aims at a deep reinforcement learning (DRL) controller for fast (< 1.5s) manipulation of a flexible tool (i.e., whip) to hit a target in 3D space. The controller consists of a DRL algorithm for optimizing joint motions, and a proportional-derivative (PD) mechanism for tracking the optimized motions. Their objective is to minimize the distance between the whip-end-tip and the target. The proposed controller was validated in a 7-DOF robot arm by comparing four DRL algorithms in the physical simulator MuJoCo. It shows that the proximal policy optimization (PPO) outperforms others by obtaining the maximum average reward. Notably, PPO can still effectively interact with the environment under sparse or even unrewarding conditions, making it a robust choice for complex and dynamic tasks. Our work provides preliminary knowledge of DRL applications to fast robotic arm control in flexible object manipulation.

关键词： Three-dimensional displays Target tracking Heuristic algorithms Programming Aerospace electronics Deep reinforcement learning Manipulators Optimization Autonomous robots

来源：评论

学校读者我要写书评

暂无评论

DenseSplat: Densifying Gaussian Splatting SLAM with Neural Radiance Prior

arXiv

引用

arXiv 2025年

作者： Li, Mingrui Liu, Shuhong Deng, Tianchen Wang, Hongyu the Department of Computer Science Dalian University of Technology Dalian China the Department of Mechano-informatics Information Science and Technology The University of Tokyo Tokyo Japan Institute of Medical Robotics Department of Automation Shanghai Jiao Tong University Key Laboratory of System Control and Information Processing Ministry of Education China

Gaussian SLAM systems excel in real-time rendering and fine-grained reconstruction compared to NeRF-based systems. However, their reliance on extensive keyframes is impractical for deployment in real-world robotic systems, which typically operate under sparse-view conditions that can result in substantial holes in the map. To address these challenges, we introduce DenseSplat, the first SLAM system that effectively combines the advantages of NeRF and 3DGS. DenseSplat utilizes sparse keyframes and NeRF priors for initializing primitives that densely populate maps and seamlessly fill gaps. It also implements geometry-aware primitive sampling and pruning strategies to manage granularity and enhance rendering efficiency. Moreover, DenseSplat integrates loop closure and bundle adjustment, significantly enhancing frame-to-frame tracking accuracy. Extensive experiments on multiple large-scale datasets demonstrate that DenseSplat achieves superior performance in tracking and mapping compared to current state-of-the-art methods. © 2025, CC BY.

关键词： Real time systems

来源：评论

学校读者我要写书评

暂无评论

UGNA-VPR: A Novel Training Paradigm for Visual Place Recognition Based on Uncertainty-Guided NeRF Augmentation

arXiv

引用

arXiv 2025年

作者： Shen, Yehui Zhang, Lei Li, Qingqiu Zhao, Xiongwei Wang, Yue Lu, Huimin Chen, Xieyuanli College of Intelligence Science and Technology National University of Defense Technology China Faculty of Robot Science and Engineering Northeastern University China School of Electronic and Information Engineering Harbin Institute of Technology Shenzhen China State Key Laboratory of Industrial Control Technology Institute of Cyber-Systems and Control Zhejiang University China SIASUN Robot& Automation Co. Ltd China School of Computer Science Fudan University China

Visual place recognition (VPR) is crucial for robots to identify previously visited locations, playing an important role in autonomous navigation in both indoor and outdoor environments. However, most existing VPR datasets are limited to single-viewpoint scenarios, leading to reduced recognition accuracy, particularly in multi-directional driving or feature-sparse scenes. Moreover, obtaining additional data to mitigate these limitations is often expensive. This paper introduces a novel training paradigm to improve the performance of existing VPR networks by enhancing multi-view diversity within current datasets through uncertainty estimation and NeRF-based data augmentation. Specifically, we initially train NeRF using the existing VPR dataset. Then, our devised self-supervised uncertainty estimation network identifies places with high uncertainty. The poses of these uncertain places are input into NeRF to generate new synthetic observations for further training of VPR networks. Additionally, we propose an improved storage method for efficient organization of augmented and original training data. We conducted extensive experiments on three datasets and tested three different VPR backbone networks. The results demonstrate that our proposed training paradigm significantly improves VPR performance by fully utilizing existing data, outperforming other training approaches. We further validated the effectiveness of our approach on self-recorded indoor and outdoor datasets, consistently demonstrating superior results. Our dataset and code have been released at https://***/nubot-nudt/UGNA-VPR. Copyright © 2025, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：