检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Baumann, Alexander Ayala, Leonardo Seidlitz, Silvia Sellner, Jan Studier-Fischer, Alexander Özdemir, Berkin Maier-Hein, Lena Ilic, Slobodan Siemens AG Munich Germany Heidelberg Germany Medical Faculty Faculty of Mathematics and Computer Science Heidelberg University Germany NCT Heidelberg Germany HIDSS4Health Heidelberg Germany Department of Urology and Urosurgery University Medical Center Mannheim Germany Department of General Visceral and Transplantation Surgery Heidelberg University Hospital Germany Division of Intelligent Systems and Robotics in Urology DKFZ Heidelberg Germany DKFZ Hector Cancer Institute University Medical Center Mannheim Germany

Spectral imaging offers promising applications across diverse domains, including medicine and urban scene understanding, and is already established as a critical modality in remote sensing. However, variability in channel dimensionality and captured wavelengths among spectral cameras impede the development of AI-driven methodologies, leading to camera-specific models with limited generalizability and inadequate cross-camera applicability. To address this bottleneck, we introduce CARL, a model for Camera-Agnostic Representation Learning across RGB, multispectral, and hyperspectral imaging modalities. To enable the conversion of a spectral image with any channel dimensionality to a camera-agnostic embedding, we introduce wavelength positional encoding and a self-attention-cross-attention mechanism to compress spectral information into learned query representations. Spectral-spatial pre-training is achieved with a novel spectral self-supervised JEPA-inspired strategy tailored to CARL. Large-scale experiments across the domains of medical imaging, autonomous driving, and satellite imaging demonstrate our model’s unique robustness to spectral heterogeneity, outperforming on datasets with simulated and real-world cross-camera spectral variations. The scalability and versatility of the proposed approach position our model as a backbone for future spectral foundation models. © 2025, CC BY.

关键词： Image coding

来源：评论

学校读者我要写书评

暂无评论

Comparison of Monocular ROS-Based Visual SLAM Methods 1

引用

7th International Conference on Interactive Collaborative robotics, ICR 2022

作者： Safarova, Liaisan Abbyasov, Bulat Tsoy, Tatyana Li, Hongbing Magid, Evgeni Laboratory of Intelligent Robotic Systems Intelligent Robotics Department Institute of Information Technology and Intelligent Systems Kazan Federal University Kremlyovskaya Str. 35 Kazan420111 Russia Department of Instrument Science and Engineering Shanghai Jiao Tong University Shanghai China Tikhonov Moscow Institute of Electronics and Mathematics HSE University Tallinn Str 34 Moscow123592 Russia

ISBN: (数字)9783031236099

ISBN: (纸本)9783031236082

Simultaneous Localization and Mapping (SLAM) is a robot navigation approach used to estimate a movement of a sensor in an unknown environment. SLAM application examples include urban search and rescue operations in high-risk environments, visual surveillance and service robotics. Compared to laser-range finders, visual sensors are more light weight, have less power consumption and provide more vast amount of environmental information. Visual SLAM methods are based on visual information only and employ monocular, stereo or RGD-D cameras as input sensors. Monocular SLAM methods are often preferred for a robotic platform when cost, energy and system weight requirements are limited. Calculation of errors and testing on ready-made datasets allow to check applicability of an algorithm in real-world scenarios with less efforts than experiments with a real robot. Verification of a sensor position and robot trajectory estimation accuracy is achieved by measuring Absolute Trajectory Error (ATE), Relative Pose Error (RPE) and Root Mean Square Error (RMSE). This article presents analysis and comparison of the most widely used Visual SLAM monocular algorithms built for Robot Operating System (ROS): ORB-SLAM2, ORB-SLAM3, DSO and LDSO. The KITTI and Euroc MAV datasets were selected to evaluate the algorithms’ performance. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Robot Operating System

来源：评论

学校读者我要写书评

暂无评论

High-precision Calibration of Camera and IMU on Manipulator for Bio-inspired Robotic System

引用

Journal of Bionic Engineering 2022年第2期19卷 299-313页

作者： Yinlong Zhang Wei Liang Sichao Zhang Xudong Yuan Xiaofang Xia Jindong Tan Zhibo Pang State Key Laboratory of Robotics Shenyang Institute of AutomationChinese Academy of SciencesShenyang 110016China Key Laboratory of Networked Control Systems Chinese Academy of SciencesShenyang 110016China Institutes for Robotics and Intelligent Manufacturing Chinese Academy of SciencesShenyang 110169China The School of Computer Science and Technology Xidian UniversityXi’an 710071China Department of Mechanical Aerospace and Biomedical EngineeringUniversity of TennesseeKonxville 37996USA Department of Automation Technology ABB Corporate Research Sweden72178 VästerasSweden

Inspired by box jellyfish that has distributed and complementary perceptive system,we seek to equip manipulator with a camera and an Inertial Measurement Unit(IMU)to perceive ego motion and surrounding unstructured *** robot perception,a reliable and high-precision calibration between camera,IMU and manipulator is a critical *** paper introduces a novel calibration ***,we seek to correlate the spatial relationship between the sensing units and manipulator in a joint ***,the manipulator moving trajectory is elaborately designed in a spiral pattern that enables full excitations on yaw-pitch-roll rotations and x-y-z translations in a repeatable and consistent *** calibration has been evaluated on our collected visual inertial-manipulator *** systematic comparisons and analysis indicate the consistency,precision and effectiveness of our proposed calibration method.

关键词： Bio-inspired robotic system Monocular camera IMU Manipulator Calibration

来源：评论

学校读者我要写书评

暂无评论

DirectShape: Direct Photometric Alignment of Shape Priors for Visual Vehicle Pose and Shape Estimation

DirectShape: Direct Photometric Alignment of Shape Priors fo...

引用

2020 IeeE International Conference on robotics and Automation, ICRA 2020

作者： Wang, Rui Yang, Nan Stuckler, Jorg Cremers, Daniel Technical University of Munich Department of Computer Science Garching bei MUnchen85748 Germany Max Planck Institute for Intelligent Systems Tubingen Tubingen72076 Germany

ISBN: (纸本)9781728173955

Scene understanding from images is a challenging problem encountered in autonomous driving. On the object level, while 2D methods have gradually evolved from computing simple bounding boxes to delivering finer grained results like instance segmentations, the 3D family is still dominated by estimating 3D bounding boxes. In this paper, we propose a novel approach to jointly infer the 3D rigid-body poses and shapes of vehicles from a stereo image pair using shape priors. Unlike previous works that geometrically align shapes to point clouds from dense stereo reconstruction, our approach works directly on images by combining a photometric and a silhouette alignment term in the energy function. An adaptive sparse point selection scheme is proposed to efficiently measure the consistency with both terms. In experiments, we show superior performance of our method on 3D pose and shape estimation over the previous geometric approach and demonstrate that our method can also be applied as a refinement step and significantly boost the performances of several state-of-the-art deep learning based 3D object detectors. All related materials and demonstration videos are available at the project page https://***/research/vslam/direct-shape. © 2020 IeeE.

关键词： Photometry

来源：评论

学校读者我要写书评

暂无评论

Extracting Strong Policies for robotics Tasks from Zero-Order Trajectory Optimizers 9

Extracting Strong Policies for Robotics Tasks from Zero-Orde...

引用

9th International Conference on Learning Representations, ICLR 2021

作者： Pinneri, Cristina Sawant, Shambhuraj Blaes, Sebastian Martius, Georg Autonomous Learning Group Max Planck Institute for Intelligent Systems Tübingen Germany Department of Computer Science ETH Zurich Switzerland Max Planck ETH Center for Learning Systems Department of Engineering Cybernetics NTNU Trondheim Norway

Solving high-dimensional, continuous robotic tasks is a challenging optimization problem. Model-based methods that rely on zero-order optimizers like the cross-entropy method (CEM) have so far shown strong performance and are considered state-of-the-art in the model-based reinforcement learning community. However, this success comes at the cost of high computational complexity, being therefore not suitable for real-time control. In this paper, we propose a technique to jointly optimize the trajectory and distill a policy, which is essential for fast execution in real robotic systems. Our method builds upon standard approaches, like guidance cost and dataset aggregation, and introduces a novel adaptive factor which prevents the optimizer from collapsing to the learner's behavior at the beginning of the training. The extracted policies reach unprecedented performance on challenging tasks like making a humanoid stand up and opening a door without reward shaping. © 2021 ICLR 2021 - 9th International Conference on Learning Representations. All rights reserved.

关键词： Real time control

来源：评论

学校读者我要写书评

暂无评论

Automatic programming of coordinate measuring machines

Automatic programming of coordinate measuring machines

引用

IeeE International Conference on robotics and Automation (ICRA)

作者： A.J. Spyridi A.A.G. Requicha Computer Science Department and Institute for Robotics and Intelligent Systems University of Southern California Los Angeles CA USA

This paper discusses a planner that generates automatically high-level programs (or plans) for inspecting mechanical parts with coordinate measuring machines. The system's input is a solid model of a toleranced part, and the output a collection of setups, probes and probe orientations associated with the surface features to be inspected. The problem is formulated as state space search. The state contains a collection of inspection plans, represented as a set of 4-tuples of the form (setup, probe, probe direction, surface feature). The initial plans may be invalid, or unnecessarily expensive. The state evolves through the application of operators, which impose geometric and optimization constraints so as to produce a valid and nearly optimal plan. The paper discusses the planner architecture and implementation, and presents experimental results.< >

关键词： Automatic programming Coordinate measuring machines Probes Inspection Robot kinematics intelligent robots State-space methods Manufacturing processes Education computer graphics

来源：评论

学校读者我要写书评

暂无评论

3D Face Tracking and Expression Inference from a 2D Sequence Using Manifold Learning

3D Face Tracking and Expression Inference from a 2D Sequence...

引用

26th IeeE Conference on computer Vision and Pattern Recognition (CVPR 2008), vol.11

作者： Wei-Kai Liao Gerard Medioni Computer Science Department Institute for Robotics and Intelligent Systems University of Southern California Los Angeles CA USA

We propose a person-dependent, manifold-based approach for modeling and tracking rigid and nonrigid 3D facial deformations from a monocular video sequence. The rigid and nonrigid motions are analyzed simultaneously in 3D, by automatically fitting and tracking a set of landmarks. We do not represent all nonrigid facial deformations as a simple complex manifold, but instead decompose them on a basis of eight 1D manifolds. Each 1D manifold is learned offline from sequences of labeled expressions, such as smile, surprise, etc. Any expression is then a linear combination of values along these 8 axes, with coefficient representing the level of activation. We experimentally verify that expressions can indeed be represented this way, and that individual manifolds are indeed 1D. The manifold dimensionality estimation, manifold learning, and manifold traversal operation are all implemented in the N-D Tensor Voting framework. Using simple local operations, this framework gives an estimate of the tangent and normal spaces at every sample, and provides excellent robustness to noise and outliers. The output of our system, besides the tracked landmarks in 3D, is a labeled annotation of the expression. We demonstrate results on a number of challenging sequences.

关键词： Deformable models Active appearance model Active shape model Face detection intelligent robots Facial animation Humans Principal component analysis computer science robotics and automation

来源：评论

学校读者我要写书评

暂无评论

The multi-dimensional quality of task requirements for dextrous robot hand control

The multi-dimensional quality of task requirements for dextr...

引用

IeeE International Conference on robotics and Automation (ICRA)

作者： H. Liu T. Iberall G.A. Bekey Institute for Robotics and Intelligent Systems Computer Science Department University of Southern California Los Angeles CA USA

The authors identify four important task requirements for dextrous robot hand control. These requirements are stability, manipulability, torquability, and radial rotatability. High-level task descriptions, supplied by the user, are refined into detailed task descriptions that can be used to drive a robot hand. A knowledge-based approach for refining a reasonable set of tasks is used to infer values for a set of task attributes, which trigger several heuristics. Those heuristics are applied using a set of metaheuristics to determine good grasp postures and poses for the task. How to use this multidimensional grasping quality in grasp mode selection and performance evaluation is shown in an industrial assembly domain.< >

关键词： Robot control intelligent robots Robotic assembly Control systems Fingers Stability Refining Service robots robotics and automation Manufacturing automation

来源：评论

学校读者我要写书评

暂无评论

3-D descriptions of buildings from an oblique view aerial image

3-D descriptions of buildings from an oblique view aerial im...

引用

International Symposium on computer Vision (ISCV)

作者： Chungan Lin R. Nevatia Institute for Robotics and Intelligent Systems Computer Science Department University of Southern California Los Angeles CA USA

In this paper we describe a method to reconstruct the 3-D descriptions of buildings from an oblique view aerial image. Oblique views present new difficulties over the more conventional nadir views that are now handled by our system. A hierarchical perceptual grouping process is used to generate 2-D roof hypotheses from fragmented linear features of the input image. Good hypotheses are selected and then verified by the corresponding shadow and walls. The 3-D description of a building is reconstructed from the roof hypothesis and the associated shadow and wall evidence. Results on several oblique view images are shown and the evaluation of the results is presented also.

关键词： Image reconstruction Shape Buildings Layout intelligent robots Image segmentation Image edge detection intelligent systems intelligent structures computer science

来源：评论

学校读者我要写书评

暂无评论

Implementing robots in hardware as a tool for integration

Implementing robots in hardware as a tool for integration

引用

IeeE International Symposium on intelligent Control (ISIC)

作者： G.S. Sukhatme J.F. Montgomery G.A. Bekey Department of Computer Science Institute for Robotics and Intelligent Systems University of Southern California Los Angeles CA USA

We suggest that integration in robotics research is facilitated and enhanced by the process of building embodied robots. It is in the translation of conceptual and simulated models into hardware that the integration takes place. Further, multiple techniques that address similar aspects of the same robotic task have to be merged to achieve good overall performance. In this paper we present two examples of successful integration and cross platform validation namely: 1. A walking and a wheeled robot used for planetary exploration (both used the same obstacle avoidance technique but the implementation was necessarily different) and, 2. Two robot helicopters (where behavior-based control used on the first robot is augmented using learning and enriched sensing on the second). The paper reports both case studies and discusses a list of lessons learned in the process of implementation.

关键词： Hardware Mobile robots Legged locomotion intelligent robots Robot sensing systems Fabrication Orbital robotics Testing Mars Infrared sensors

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：