检索结果-内蒙古大学图书馆

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Chen, Hui Liu, Yanbin Ma, Yongqiang Zheng, Nanning Yu, Xin National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center of Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University China Auckland University of Technology New Zealand The University of Queensland Australia

Pre-trained vision-language models (VLMs) such as CLIP have shown excellent performance for zero-shot classification. Based on CLIP, recent methods design various learnable prompts to evaluate the zero-shot generalization capability on a base-to-novel setting. This setting assumes test samples are already divided into either base or novel classes, limiting its application to realistic scenarios. In this paper, we focus on a more challenging and practical setting: generalized zero-shot learning (GZSL), i.e., testing with no information about the base/novel division. To address this challenging zero-shot problem, we introduce two unique designs that enable us to classify an image without the need of knowing whether it comes from seen or unseen classes. Firstly, most existing methods only adopt a single latent space to align visual and linguistic features, which has a limited ability to represent complex visual-linguistic patterns, especially for fine-grained tasks. Instead, we propose a dual-space feature alignment module that effectively augments the latent space with a novel attribute space induced by a well-devised attribute reservoir. In particular, the attribute reservoir consists of a static vocabulary and learnable tokens complementing each other for flexible control over feature granularity. Secondly, finetuning CLIP models (e.g., prompt learning) on seen base classes usually sacrifices the model's original generalization capability on unseen novel classes. To mitigate this issue, we present a new topology-preserving objective that can enforce feature topology structures of the combined base and novel classes to resemble the topology of CLIP. In this manner, our model will inherit the generalization ability of CLIP through maintaining the pairwise class angles in the attribute space. Extensive experiments on twelve object recognition datasets demonstrate that our model, termed Topology-Preserving Reservoir (TPR), outperforms strong baselines including both pro

关键词：

来源：评论

学校读者我要写书评

暂无评论

Spectral Aggregation Cross-Square Transformer for Hyperspectral Image Denoising 27th

Spectral Aggregation Cross-Square Transformer for Hyperspect...

引用

27th International Conference on Pattern Recognition, ICPR 2024

作者： Liu, Yang Ji, Yantao Xiao, Jiahua Guo, Yu Jiang, Peilin Yang, Haiwei Wang, Fei School of Software Engineering Xi’an Jiaotong University Shaanxi Xi’an710049 China National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center for Visual Information and Applications and Institute of Artificial Intelligence and Robotics Xi’an Jiaotong University Shaanxi Xi’an710049 China

ISBN: (纸本)9783031783531

Hyperspectral image(HSI) denoising addresses noise impact during image acquisition. Transformers have gained notable prominence in the field of denoising, but their quadratic self-attention complexity poses computational challenges, hindering global information processing. Classical window-based self-attention limits non-local information flow, hampering large-scale object capture in HSI. Furthermore, spectral variations among neighboring bands introduce redundancy, which burdens the model and diminishes token variability, resulting in over-smoothing in the attention map. To address these issues, we propose a novel method, named Spectral Aggregation Cross-Square Transformer(SACT). We introduce a cross-square self-attention mechanism to enhance information exchange between windows, capturing long-range dependencies within spatial intra-spectrum from multiple perspectives. Spectrally, it extends the attention region horizontally, surrounding, and vertically, exploring omnidirectional spatial correlations among different receptive windows. Additionally, a spatial-spectral aggregation self-attention module is designed to capture global contextual dependencies across spatial and spectral dimensions, reducing spectral redundancy computation. Our method has evaluated synthetic and real hyperspectral datasets and shows SACT’s effectiveness in enhancing both quantitative and qualitative HSI denoising performance. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Image denoising

来源：评论

学校读者我要写书评

暂无评论

Correcting of the unexpected localization measurement for indoor automatic mobile robot transportation based on a neural network

引用

Transportation Safety and Environment 2024年第2期6卷 24-35页

作者： Jiahao Huang Steffen Jung inger Hui Liu Kerstin Thurow Center for Life Science Automation(CELISCA) University of RostockRostock 18119Germany Institute of Automation University of RostockRostock 18119Germany Institute of Artificial Intelligence&Robotics(IAIR) School of Traffic&Transportation EngineeringCentral South UniversityChangsha 410075HunanChina

The increasing use of mobile robots in laboratory settings has led to a higher degree of laboratory ***,when mobile robots move in laboratory environments,mechanical errors,environmental disturbances and signal interruptions are *** can compromise the accuracy of the robot’s localization,which is crucial for the safety of staff,robots and the laboratory.A novel time-series predicting model based on the data processing method is proposed to handle the unexpected localization measurement of mobile robots in laboratory *** proposed model serves as an auxiliary localization system that can accurately correct unexpected localization errors by relying solely on the historical data of mobile *** experimental results demonstrate the effectiveness of this proposed method.

关键词： mobile robots laboratory automation indoor localization neural network

来源：评论

学校读者我要写书评

暂无评论

Laser-powered multi-UAV URLLC Systems: Reliability and scheduling performance analysis

引用

IEEE Transactions on Vehicular Technology 2023年第11期72卷 14615-14630页

作者： Singh, Keshav Raut, Prasanna Sharma, Prabhat Kumar Li, Chih-Peng National Sun Yat-sen University Institute of Communications Engineering Kaohsiung80424 Taiwan Technology Innovation Institute Autonomous Robotics Research Centre Abu Dhabi51133 United Arab Emirates Visvesvaraya National Institute of Technology Department of Electronics and Communication Engineering Maharashtra Nagpur440010 India

This article investigates the multi-unmanned aerial vehicle (UAV) communication system with multiple full-duplex (FD) mobile users under ultra-reliable low-latency communication (URLLC) constraints. The energy-constrained UAVs are powered with the help of laser beams sourced by the laser beam directors (LBDs) which are randomly deployed with Poisson point Process (PPP). The received power from LBDs is harvested at UAVs using the power-splitting energy harvesting (EH) architecture to improve its lifespan. Moreover, the received power at the UAV is modeled using the free-space optical (FSO) range equation, and is randomly affected by atmospheric turbulence and misalignment errors. With multi-user diversity scenario, we provide the mathematical framework for three user scheduling schemes, namely, random (RND) scheduling, absolute (ABS) signal-To-interference-And-noise-ratio (SINR) based scheduling and normalized (NRM) SINR scheduling schemes, wherein the UAV communicates with a single scheduled user in a given slot due to its limited abilities. The expressions for the outage probability, coverage probability and the block error rate (BLER) are derived for all the scheduling schemes under consideration. The effects of various parameters such as shadowing severity, URLLC parameters, EH, laser beam's angular spread and imperfect CSI are analyzed. The derived expressions are validated through Monte-Carlo (MC) simulations. © 1967-2012 IEEE.

关键词： Signal to noise ratio

来源：评论

学校读者我要写书评

暂无评论

An Efficient Sampling and Search Method Based on Risk Framework in Highly Dynamic Scenarios

An Efficient Sampling and Search Method Based on Risk Framew...

引用

2024 China Automation Congress, CAC 2024

作者： Shu, Xinyu Liu, Jianyi Chen, Ming National Key Laboratory of Human-Machine Hybrid Augmented Intelligence National Engineering Research Center of Visual Information and Applications Institute of Artificial Intelligence and Robotics Xi’an Jiaotong University Shaanxi 710049 China Cainiao Smart Logistics Network Limited China

ISBN: (纸本)9798350368604

It is still a challenge to quickly plan a high-quality path that is collision-free and kinematically feasible in a highly dynamic environment with dense dynamic obstacles. Traditional risk-based motion planning algorithms solve the planning problem in the environment of human-machine coexistence by considering the risk of collision with dynamic and static obstacles. Although these algorithms solve the security problems of planning, they have problems such as too many iterations and low quality of planning paths. Therefore, combined with the Artificial Potential Field method (APF), this paper introduces an efficient sampling method into the risk-based motion planning algorithm to guide the growth of random trees to the feasible region, effectively reduce the redundant nodes of random trees, and improve the quality of planning trajectories. Finally, simulation experiments are carried out in different highly dynamic hybrid scenarios, and the effectiveness of the proposed method is verified by comparing with Risk-RRT and Multi-Risk-RRT. © 2024 IEEE.

关键词：

来源：评论

学校读者我要写书评

暂无评论

PolyMerge: A Novel Technique aimed at Dynamic HD Map Updates Leveraging Polylines 21

PolyMerge: A Novel Technique aimed at Dynamic HD Map Updates...

引用

21st International Conference on Advanced robotics, ICAR 2023

作者： Sayed, Mohamed Perminov, Stepan Tsetserukou, Dzmitry Center for Digital Engineering Skolkovo Institute of Science and Technology Intelligent Space Robotics Laboratory Moscow Russia Llc IntegraNT Moscow Russia

ISBN: (纸本)9798350342291

Currently, High-Definition (HD) maps are a pre- requisite for the stable operation of autonomous vehicles. Such maps contain information about all static road objects for the vehicle to consider during navigation, such as road edges, road lanes, crosswalks, and etc. To generate such an HD map, current approaches need to process pre-recorded environment data obtained from onboard sensors. However, recording such a dataset often requires a lot of time and effort. In addition, every time actual road environments are changed, a new dataset should be recorded to generate a relevant HD map. This paper addresses a novel approach that allows to continuously generate or update the HD map using onboard sensor data. When there is no need to pre-record the dataset, updating the HD map can be run in parallel with the main autonomous vehicle navigation pipeline. The proposed approach utilizes the VectorMapNet framework to generate vector road object instances from a sensor data scan. The PolyMerge technique is aimed to merge new instances into previous ones, mitigating detection errors and, therefore, generating or updating the HD map. The performance of the algorithm was confirmed by comparison with ground truth on the NuScenes dataset. Experimental results showed that the mean error for different levels of environment complexity was comparable to the VectorMapNet single instance error. © 2023 IEEE.

关键词： Errors

来源：评论

学校读者我要写书评

暂无评论

Trash Detection in Water Bodies Using Yolo With Explainable AI Insight

Trash Detection in Water Bodies Using Yolo With Explainable ...

引用

International Conference on robotics and Artificial Intelligence (ICRAI)

作者： Shafaq Rasheed Amal Mirza Muhammad Saad Saeed Muhammad Haroon Yousaf Department of Computer Engineering University of Engineering and Technology Taxila Pakistan Swarm Robotics Lab National Center of Robotics and Automation Taxila Pakistan

ISBN: (数字)9798350355703

ISBN: (纸本)9798350355710

Waste disposal into water bodies is a serious concern for environmental engineers, often resulting in urban flooding, soil degradation in agricultural areas, and freshwater pollution. Additionally, trash accumulation in water bodies threatens ma-rine life by introducing contaminants through storm drains, rivers, and other channels. Furthermore, the unique character-istics of underwater environments, including high noise levels, low visibility, and color variation, pose challenges for trash detection. Thus, addressing this issue is crucial for environmental sustainability. The work focuses on both floating and submerged trash in water bodies. Typically, the detection of trash on water bodies is impacted by environmental variability, diverse appearance, and background clutter. These conditions also make usual object detection techniques less effective. Deep learning methods have shown promise in handling challenging visual tasks, therefore researchers have turned to them to get over these challenges. In this work, we focus on applying different versions of YOLO to the problem of water trash detection. To assess these models' performance, we use four different datasets: Trash Flow rate Dataset, DeepTrash, FloW, and TrashICRA. Each dataset represents various trash types under different conditions such as underwater trash and floating trash. Our study attempts to identify the best method suitable for the detection of aquatic waste. Our results highlight the effectiveness of YOLO variants in identifying the presence of trash in water ecosystems. By leveraging Grad-CAM we gain insights into the interpretability and transparency of deep learning models thus enabling Explainable AI. This approach will also contribute to the understanding and trustworthiness of artificial systems deployed for trash detection applications by providing a transparent view of decision making process of deep learning models.

关键词： Deep learning YOLO Visualization Explainable AI Storms Biological system modeling Soil Water pollution Waste disposal robotics and automation

来源：评论

学校读者我要写书评

暂无评论

Variable Stiffness Mechanism for Single-Joint Lower Limb Wearable Exoskeleton: A Review 6

Variable Stiffness Mechanism for Single-Joint Lower Limb Wea...

引用

6th International Conference on Automation, Control and Robots, ICACR 2022

作者： Mao, Zhengzhong Wang, Chunlei Rui, Yuefeng Huang, Xiandao Zhang, Zhenyu Li, Yunji Yang, Ya The 21st Research Institute of Cetc Robotics Engineering Center Shanghai China

ISBN: (数字)9781665498739

ISBN: (纸本)9781665498739

Lower limb wearable exoskeleton can support the rehabilitation of the patient and offer assistance of the healthy human. But the traditional rigid exoskeleton mechanism is not inherently compliant, resulting in human-machine collision, lacking safety and environmental adaptability. The variable stiffness mechanism (VSM) is proved can improve inherent compliance of the exoskeleton, with the advantages of wide range of stiffness adjustment, considerable control bandwidth, high flexibility and security. However, the application of VSMs in wearable exoskeletons is still in the laboratory stage with limited commercial application cases, as designing a comfortable, compact VSM that takes into account compliance, safety, and stiffness adjustment capability is a major challenge. A detailed review of existing VSMs is provided for different lower extremity joints including hip, knee, and ankle joints to accelerate the application of VSM in exoskeleton. The VSMs were classified, summarized and compared according to active and passive assistance. The review focused on the structures and interface attachment components for the mechanical compliance of the VSMs. For passive VSM, more attention should be paid to the bio-inspired compact design, strictly following the load pattern of the lower limb muscles, and flexible adjustment of the stiffness. For active VSM, it tends to combine flexible force transmission systems with novel stiffness adjustment methods. In the future, developing active-passive combined exoskeletons is bounded to be a new trend. © 2022 IEEE.

关键词： Stiffness

来源：评论

学校读者我要写书评

暂无评论

Deep Reinforcement Learning for Adaptive Job Shop Scheduling in Robotic Cells: Handling Disruptions Effectively 11

Deep Reinforcement Learning for Adaptive Job Shop Scheduling...

引用

11th International Conference on Mechatronics and robotics engineering, ICMRE 2025

作者： Miller, Eddi Schmitt, Anna-Maria Kaupp, Tobias Schiffler, Andreas Schmitt, Jan Institute of Digital Engineering Technical University of Applied Sciences Würzburg-Schweinfurt Schweinfurt Germany Center for Robotics Technical University of Applied Sciences Würzburg-Schweinfurt Schweinfurt Germany

ISBN: (纸本)9798331509293

In contemporary manufacturing, efficient operation of robotic cells is critical and Job Shop Scheduling (JSS) faces challenges such as machine failures and shifting priorities. Deep Reinforcement Learning (DRL) addresses these challenges by enabling adaptive scheduling. DRL empowers robotic cells to dynamically optimize schedules which can increase resilience drastically. Through simulations and real-world use, DRL-based approaches outperform traditional methods, improving adaptability and efficiency. Integrating DRL into scheduling processes enables better navigation of uncertainties, enhancing productivity and competitiveness in modern manufacturing. This contribution focuses on these problems and provides a DRL approach for robotic cells to dynamically schedule production orders when facing random tool breakdowns. Using the e-greedy policy, it allows the agent to explore the environment and choose different actions to generate rewards. Apart from reacting to random machine failures, the provided model enables collaborative robotic processes to further shorten the makespan and increase the utilization of industrial robots paired with machine tools. © 2025 IEEE.

关键词： Industrial robots

来源：评论

学校读者我要写书评

暂无评论

Leakage Source Localization Based on Improved-Infotaxis 16th

Leakage Source Localization Based on Improved-Infotaxis

引用

16th International Conference on Modelling, Identification and Control, ICMIC 2024

作者： Tao, Xiang Cheng, Lei Yao, Tianxiao Engineering Research Center for Metallurgical Automation and Measurement Technology of Ministry of Eduction Wuhan430081 China Institute of Robotics and Intelligent Systems Wuhan University of Science and Technology Wuhan430081 China

ISBN: (纸本)9789819617760

In recent years, with the rapid development of the chemical industry, the risks in this field are also increasing, in which the leakage of hazardous gases can cause serious harm to both people and the environment, so it is crucial to locate the leakage source quickly and accurately. For pollution source traceability, the traditional gradient algorithm is prone to fall into local optimization, and the infotaxis algorithm is biased towards exploration, which is prone to waste time cost and control cost. We propose a new strategy to balance the exploration and utilization in the search process by studying the effects of multiple algorithms and setting different experimental parameters (gas release rate and wind speed) on the overall performance, and the experimental results show that the algorithm has a strong performance under different environmental variables with strong robustness. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Chemical industry

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：