检索结果-内蒙古大学图书馆

IEEE International Conference on robotics and Automation (ICRA)

作者： Jiahao Ruan Li He Yisheng Guan Hong Zhang School of Electromechanical Engineering Guangdong University of Technology Guangzhou China Department of Electrical and Electronic Engineering Shenzhen Key - Laboratory of Robotics and Computer Vision Southern University of Science and Technology Shenzhen China

Visual relocalization is a fundamental problem in computer vision and robotics. Recently, regression-based methods become popular and they can be categorized into two classes: absolute pose regression and scene coordinate regression. In this work, we present a combined regression network that jointly learns scene coordinate regression and absolute pose regression for single-image visual relocalization. The proposed network composes of a feature encoder and two regression branches with uncertainty modeling. In particular, we design a deep feature conditioning module, aiming at propagating the coarse pose information in absolute pose regression to inform the predictions in scene coordinate regression. The proposed network is trained in an end-to-end fashion to learn both regression tasks. Moreover, we propose an uncertainty-driven RANSAC algorithm that incorporates the predicted scene coordinates and their uncertainties to solve the camera pose during inference. To the best of our knowledge, this work is the first to combine scene coordinate regression and pose regression in a hierarchical framework for visual relocalization. Experiments on indoor and outdoor benchmarks demonstrate the effectiveness and the superiority of the proposed method over the state-of-the-art methods.

关键词：

来源：评论

学校读者我要写书评

暂无评论

FoundationGrasp: Generalizable Task-Oriented Grasping with Foundation Models

arXiv

引用

arXiv 2024年

作者： Tang, Chao Huang, Dehao Dong, Wenlong Xu, Ruinian Zhang, Hong Shenzhen Key Laboratory of Robotics and Computer Vision Southern University of Science and Technology Shenzhen China Department of Electronic and Electrical Engineering Southern University of Science and Technology Shenzhen China Institute for Robotics and Intelligent Machines Georgia Institute of Technology Atlanta United States

Task-oriented grasping (TOG), which refers to synthesizing grasps on an object that are configurationally compatible with the downstream manipulation task, is the first milestone towards tool manipulation. Analogous to the activation of two brain regions responsible for semantic and geometric reasoning during cognitive processes, modeling the intricate relationship between objects, tasks, and grasps necessitates rich semantic and geometric prior knowledge about these elements. Existing methods typically restrict the prior knowledge to a closed-set scope, limiting their generalization to novel objects and tasks out of the training set. To address such a limitation, we propose FoundationGrasp, a foundation model-based TOG framework that leverages the open-ended knowledge from foundation models to learn generalizable TOG skills. Extensive experiments are conducted on the contributed Language and vision Augmented TaskGrasp (LaViA-TaskGrasp) dataset, demonstrating the superiority of FoundationGrasp over existing methods when generalizing to novel object instances, object classes, and tasks out of the training set. Furthermore, the effectiveness of FoundationGrasp is validated in real-robot grasping and manipulation experiments on a 7-DoF robotic arm. Our code, data, appendix, and video are publicly available at https://***/view/foundationgrasp. Copyright © 2024, The Authors. All rights reserved.

关键词： Robotic arms

来源：评论

学校读者我要写书评

暂无评论

Long-Term Invariant Local Features via Implicit Cross-Domain Correspondences

arXiv

引用

arXiv 2023年

作者： Pataki, Zador Altillawi, Mohammad Kanakis, Menelaos Pautrat, Rémi Shen, Fengyi Liu, Ziyuan Van Gool, Luc Pollefeys, Marc The Computer Vision and Geometry Lab Department of Computer Science ETH Zurich Switzerland The Computer Vision Center CVC-Barcelona The Intelligent Robotics Cloud Technology lab of Huawei-Munich Germany The Computer Vision Lab Department Electrical Engineering ETH Zurich Switzerland The Intelligent Robotics Cloud Technology lab of Huawei-Munich Germany The Intelligent Robotics Cloud Technology lab of Huawei-Munich Germany The Center for Processing Speech and Images KU Leuven The Computer Vision Lab ETH Zurich Switzerland

Modern learning-based visual feature extraction networks perform well in intra-domain localization, however, their performance significantly declines when image pairs are captured across long-term visual domain variations, such as different seasonal and daytime variations. In this paper, our first contribution is a benchmark to investigate the performance impact of long-term variations on visual localization. We conduct a thorough analysis of the performance of current state-of-the-art feature extraction networks under various domain changes and find a significant performance gap between intra- and cross-domain localization. We investigate different methods to close this gap by improving the supervision of modern feature extractor networks. We propose a novel data-centric method, Implicit Cross-Domain Correspondences (iCDC). iCDC represents the same environment with multiple Neural Radiance Fields, each fitting the scene under individual visual domains. It utilizes the underlying 3D representations to generate accurate correspondences across different long-term visual conditions. Our proposed method enhances cross-domain localization performance, significantly reducing the performance gap. When evaluated on popular long-term localization benchmarks, our trained networks consistently outperform existing methods. This work serves as a substantial stride toward more robust visual localization pipelines for long-term deployments, and opens up research avenues in the development of long-term invariant descriptors. Copyright © 2023, The Authors. All rights reserved.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Collision Avoidance for Convex Primitives via Differentiable Optimization Based High-Order Control Barrier Functions

arXiv

引用

arXiv 2024年

作者： Wei, Shiqing Khorrambakht, Rooholla Krishnamurthy, Prashanth Gonçalves, Vinicius Mariano Khorrami, Farshad Control/Robotics Research Laboratory Electrical and Computer Engineering Department Tandon School of Engineering New York University BrooklynNY11201 United States MG Belo Horizonte31270-901 Brazil

Ensuring the safety of dynamical systems is crucial, where collision avoidance is a primary concern. Recently, control barrier functions (CBFs) have emerged as an effective method to integrate safety constraints into control synthesis through optimization techniques. However, challenges persist when dealing with convex primitives and tasks requiring torque control, as well as the occurrence of unintended equilibria. This work addresses these challenges by introducing a high-order CBF (HOCBF) framework for collision avoidance among convex primitives. We transform nonconvex safety constraints into linear constraints by differentiable optimization and prove the high-order continuous differentiability. Then, we employ HOCBFs to accommodate torque control, enabling tasks involving forces or high dynamics. Additionally, we analyze the issue of spurious equilibria in high-order cases and propose a circulation mechanism to prevent the undesired equilibria on the boundary of the safe set. Finally, we validate our framework with three experiments on the Franka research 3 robotic manipulator, demonstrating successful collision avoidance and the efficacy of the circulation mechanism. © 2024, CC BY.

关键词： Quadratic programming

来源：评论

学校读者我要写书评

暂无评论

A Joint Tracking System: Robot is Online to Access Surveillance Views

A Joint Tracking System: Robot is Online to Access Surveilla...

引用

IEEE International Conference on robotics and Biomimetics

作者： Zheyuan Lin Shanshan Ji Wen Wang Mengjie Qin Rong Yang Minhong Wan Jason Gu Te Li Chunlong Zhang Research Center for Intelligent Robotics Zhejiang Laboratory Research Institute of Interdisciplinary Innovation Hangzhou China Zhejiang Engineering Research Center for Intelligent Robotics Hangzhou China Department of Electrical and Computer Engineering Dalhousie University Halifax Canada

The application of robots in social life, equipped with sensors and actuators and embedded with AI, assists people in all aspects. However, the first perspective of the robot horizon is heavily constrained, which weakens its performance. A joint tracking system is designed and built to deal with this, by integrating a surveillance system with the robot visual, providing a third perspective. This system takes one horizontal view and two top views from various directions as inputs and matches a person among the frames and in time sequence. In order to deal with the identity match with a huge visual feature gap, a special dataset is collected, simultaneously labeling identities from a mobile robot perspective and multiple indoor static surveillance monitors. The experiment shows that such match is a task worth exploring that can be better handled by training on our dataset than existing open source Re-identification (Re-id) datasets. Moreover, in the real scenario, this system improves the performance on issues like in and out of the robot’s field of vision and heavy occlusion by people or objects.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Novel Stiffness Programmable Method for Environment-adaptive Crawling Robot

A Novel Stiffness Programmable Method for Environment-adapti...

引用

2022 IEEE International Conference on robotics and Biomimetics, ROBIO 2022

作者： Luo, Yudong Zhao, Na Shen, Yantao Dalian Maritime University Morphing Robotics Laboratory Department of Computer Science and Technology Liaoning Dalian116021 China University of Nevada Department of Electrical and Biomedical Engineering RenoNV89557 United States

ISBN: (纸本)9781665481090

Benefitting from the adjustable body stiffness, crawling animals such as earthworms can crawl in soil with different hardnesses. To address one of the main challenges of variable stiffness in crawling robots, this paper presents a new concept of tuning stiffness characteristics based on the mechanical stiffening method, constructs the stiffness model of the robot, and performs a proof-of-concept based on a custom-built crawling robot prototype. The robot is composed of a group of specially designed scissor elements, which allows it to achieve bidirectional continuous deformation, thereby enabling continuous stiffness regulation. The stiffness can be adjusted simply by programming the input to the actuators or modi-fying the number of scissor elements, making it easy control, fast response, and high energy efficiency. Then, the stiffness mathematical model of the novel crawling robot segment is established, and the physical parameters are excavated on the basis of numerous numerical simulations. Finally, the 3D-printed prototype of the robot is constructed for experiments, and the results demonstrate the feasibility of tuning the stiffness characteristics based on the proposed scissor element enabled mechanism, which provides a new perspective on the way to change the stiffness of the crawling robot. © 2022 IEEE.

关键词： Stiffness

来源：评论

学校读者我要写书评

暂无评论

Virtual Reality Solutions for Troubleshooting Autonomous Industrial Mobile Robots

Virtual Reality Solutions for Troubleshooting Autonomous Ind...

引用

International Symposium on INFOTEH-JAHORINA (INFOTEH)

作者： Cihat Talat Akpinar Özcan Koşar Akif Durdu Computer Engineering Research and Development Elfatek Elektronik Ltd. Co. Konya Turkey Department of Electrical and Electronics Engineering Robotics Automation Control Laboratory (RAC-LAB) Konya Technical University Konya Turkey

ISBN: (数字)9798331515799

ISBN: (纸本)9798331515805

This study aims to develop a virtual reality (VR) application to improve the fault detection and troubleshooting processes of autonomous industrial mobile robots. The complex systems and operating environments of autonomous robots make fault detection challenging for users, requiring effective and user-friendly solutions to overcome these difficulties. In this context, the Robot Operating System (ROS) was utilized to detect errors in robots and record these errors in a database. Thanks to the modular structure and robust communication infrastructure offered by ROS, faults in the robots were monitored and analyzed through ROS *** study demonstrates the effective use of VR technology in the maintenance and repair processes of autonomous industrial mobile robots. By integrating ROS-based data collection and fault detection into a VR environment for user training, robot operators are enabled to troubleshoot faults more quickly and effectively. In this way, it is aimed to achieve time and cost savings in industrial processes.

关键词： Training Service robots Databases Fault detection Operating systems Virtual reality Maintenance engineering Maintenance Mobile robots Monitoring

来源：评论

学校读者我要写书评

暂无评论

Enhancing the Performance of Machine Learning Classification Models

Enhancing the Performance of Machine Learning Classification...

引用

International Symposium on INFOTEH-JAHORINA (INFOTEH)

ISBN: (数字)9798331515799

ISBN: (纸本)9798331515805

This study aims to compare and improve the performance of four different machine learning algorithms (Naive Bayes, Multi-Layer Perceptron (MLP), Decision Trees, and Support Vector Machines (SVM)) in classification problems. The analyses emphasize the impact of data preprocessing steps and hyperparameter optimization on model performance. As part of the data preprocessing, missing values were imputed, categorical data were transformed into numerical data, and normalization procedures were applied. It was observed that normalization significantly enhanced the performance of the MLP and SVM algorithms in particular. Furthermore, additional improvements in accuracy rates were achieved through hyperparameter optimization. Naive Bayes and Decision Trees were found to exhibit stable performance regardless of data scaling. This study demonstrates that proper data preprocessing and model selection can significantly enhance algorithm performance in classification problems.

关键词： Support vector machines Accuracy Machine learning algorithms Data preprocessing Hyperparameter optimization Data models Bayes methods Classification algorithms Numerical models Decision trees

来源：评论

学校读者我要写书评

暂无评论

Discharge image reconstruction and frequency domain analysis based on event data

引用

High Voltage 2024年第6期9卷 1195-1201页

作者： Quan Yuan Lei Deng Hao Guo Qishen Lyu Xin Zhang Jibin Wu Yu Deng Hongyang Zhou Xilin Wang Zhidong Jia Engineering Laboratory of Power Equipment Reliability in Complicated Coastal Environments Tsinghua Shenzhen International Graduate SchoolTsinghua UniversityShenzhenGuangdongChina Department of Precision Instrument Center for Brain Inspired Computing ResearchTsinghua UniversityBeijingChina College of Information and Computer Taiyuan University of TechnologyTaiyuanChina Shenzhen Power Supply Corporation China Southern Power Grid CorporationShenzhenChina Department of Computing The Hong Kong Polytechnic UniversityHong KongChina China Electric Power Research Institute BeijingChina School of Electrical Engineering and Automation Xiamen University of TechnologyXiamenChina

Optical image method has been the earliest and most used direct method for observing gas ***,research on gas discharge monitoring based on visible light mainly relies on high-speed cameras,but the large size,significant data storage requirements,and susceptibility to interference from complex backgrounds and lighting conditions limit their further *** vision sensing(DVS)technology is a neuromorphic sensing technology that asynchronously measures the luminance changes at each *** offers advantages such as a large dynamic range(>120 dB),high temporal resolution(up to 1µs),and small data volume(MB level).In this study,dynamic vision sensing technology was employed to monitor both 30 mm short-gaps and 1080 mm long-gaps discharge processes *** study developed the CountImage encoding method for event data and conducted image reconstruction,time-domain analysis,and frequency-domain charac-teristic analysis based on the event *** results show that the event-reconstructed images are highly consistent with the high-speed camera images,and the arc develop-ment process and its path can also be clearly ***,this study discovered a correlation between the electrical characteristics and event information during the discharge *** the time domain,the duration of the maximum DVS event count closely matches the duration during which the voltage drops to zero during *** the frequency domain,the Pearson correlation coefficient between the event stream spectrum and the voltage signal spectrum is greater than *** the maximum number of brightening events(ONmax)and the maximum number of darkening events(OFFmax)are positively correlated with the voltage applied between the *** study demonstrates that,compared to the GB/s data rate of high-speed cameras,this approach can record the discharge process and accurately reconstruct the discharge process,arc morphology,and discharge path at M

关键词： image analysis holds

来源：评论

学校读者我要写书评

暂无评论

MSMB-GCN: Multi-scale Multi-branch Fusion Graph Convolutional Networks for 3D Human Pose Estimation

MSMB-GCN: Multi-scale Multi-branch Fusion Graph Convolutiona...

引用

IEEE International Conference on robotics and Biomimetics

作者： Shanshan Ji Qiwei Meng Wen Wang Zheyuan Lin Te Li Minhong Wan Chunlong Zhang Jason Gu Research Center for Intelligent Robotics Research Institute of Interdisciplinary Innovation Zhejiang Laboratory Hangzhou China Department of Electrical and Computer Engineering Dalhousie University Halifax NS Canada

In human-robot interaction (HRI), human pose estimation is a necessary technology for the robot to perceive the dynamic environment and make interactive actions. Recently, graph convolutional networks (GCNs) have been increasingly used for 2D to 3D pose estimation tasks since the skeleton topologies can be viewed as graph structures. In this paper, we propose a novel graph convolutional network architecture, Multi-scale Multi-branch Fusion Graph Convolutional Networks (MSMB-GCN), for 3D Human Pose Estimation(3D HPE) task. The proposed model consists of multiple GCN blocks with a multi-branch architecture. This multi-branch architecture enables the model to get multi-scale features for human skeletal representations. The group of GCN blocks, which has strong multi-level feature extraction capabilities, allows the model to learn global and local features, lower-level and higher-level features. Experiment results on the HumanPose benchmark demonstrate that our model outperforms the state-of-the-art and ablation studies validate the effectiveness of our approach.

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：