检索结果-内蒙古大学图书馆

Learning Embeddings with Centroid Triplet Loss for Object Identification in Robotic Grasping

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Gouda, Anas Schwarz, Max Reining, Christopher Behnke, Sven Kirchheim, Alice TU Dortmund Germany Autonomous Intelligent Systems - Computer Science VI Center for Robotics University of Bonn Germany Lamarr Institute for Machine Learning and Artificial Intelligence Germany Fraunhofer IML Germany

Foundation models are a strong trend in deep learning and computer vision. These models serve as a base for applications as they require minor or no further fine-tuning by developers to integrate into their applications. Foundation models for zero-shot object segmentation such as Segment Anything (SAM) output segmentation masks from images without any further object information. When they are followed in a pipeline by an object identification model, they can perform object detection without training. Here, we focus on training such an object identification model. A crucial practical aspect for an object identification model is to be flexible in input size (number of input images). As object identification is an image retrieval problem, a suitable method should handle multi-query multigallery situations without constraining the number of input images (e.g. by having fixed-size aggregation layers). The key solution to train such a model is the centroid triplet loss (CTL), which aggregates image features to their centroids. CTL yields high accuracy, avoids misleading training signals and keeps the model input size flexible. In our experiments, we establish a new state of the art on the ArmBench object identification task, which shows general applicability of our model. We furthermore demonstrate an integrated unseen object detection pipeline on the challenging HOPE dataset, which requires fine-grained detection. There, our pipeline matches and surpasses related methods which have been trained on dataset-specific data. Code and pretrained models are available. Copyright © 2024, The Authors. All rights reserved.

关键词： Image segmentation

Directed Exploration in Reinforcement Learning from Linear Temporal Logic

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Bagatella, Marco Krause, Andreas Martius, Georg Department of Computer Science ETH Zürich Zürich Switzerland Max Planck Institute for Intelligent Systems Tübingen Germany

Linear temporal logic (LTL) is a powerful language for task specification in reinforcement learning, as it allows describing objectives beyond the expressivity of conventional discounted return formulations. Nonetheless, recent works have shown that LTL formulas can be translated into a variable rewarding and discounting scheme, whose optimization produces a policy maximizing a lower bound on the probability of formula satisfaction. However, the synthesized reward signal remains fundamentally sparse, making exploration challenging. We aim to overcome this limitation, which can prevent current algorithms from scaling beyond low-dimensional, short-horizon problems. We show how better exploration can be achieved by further leveraging the LTL specification and casting its corresponding Limit Deterministic Büchi Automaton (LDBA) as a Markov reward process, thus enabling a form of high-level value estimation. By taking a Bayesian perspective over LDBA dynamics and proposing a suitable prior distribution, we show that the values estimated through this procedure can be treated as a shaping potential and mapped to informative intrinsic rewards. Empirically, we demonstrate applications of our method from tabular settings to high-dimensional continuous systems, which have so far represented a significant challenge for LTL-based reinforcement learning algorithms. Copyright © 2024, The Authors. All rights reserved.

关键词： Temporal logic

Interaction-Aware Maneuver Intention Prediction Using Gated Mixture-of-Experts Attention Mechanism

学校读者我要写书评

暂无评论

Interaction-Aware Maneuver Intention Prediction Using Gated ...

International Conference on intelligent Transportation

作者： Iago Pachêco Gomes Cristiano Premebida Denis Fernando Wolf Institute of Mathematics and Computer Science University of São Paulo Brazil Department of Electrical and Computer Engineering Institute of Systems and Robotics University of Coimbra Portugal

This paper addresses the task of predicting the behavior of traffic participants, which involves complexities such as road geometry and agent interactions. To overcome these challenges, this paper presents a novel framework called AIMP (Attention-based Interaction-aware Maneuver Prediction). AIMP utilizes interaction graphs to extract intricate interaction features from traffic scenes. The framework incorporates a Gated Mixture-of-Experts Attention Mechanism, which combines information from road geometry, interaction patterns, and motion dynamics. This fusion process also considers prior maneuver intention estimations, enhancing both explainability and informativeness. Experimental results highlight a performance enhancement (approximately 2% ∼ 9% of accuracy) of the proposed AIMP framework compared to alternative fusion methods.

关键词：

FoundationGrasp: Generalizable Task-Oriented Grasping with Foundation Models

学校读者我要写书评

暂无评论

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

A Probabilistic Model behind Self-Supervised Learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Bizeul, Alice Schölkopf, Bernhard Allen, Carl Department of Computer Science ETH AI Center ETH Zurich Switzerland Max Planck Institute for Intelligent Systems Tübingen Germany

In self-supervised learning (SSL), representations are learned via an auxiliary task without annotated labels. A common task is to classify augmentations or different modalities of the data, which share semantic content (e.g. an object in an image) but differ in style (e.g. the object’s location). Many approaches to self-supervised learning have been proposed, e.g. SimCLR, CLIP, and DINO, which have recently gained much attention for their representations achieving downstream performance comparable to supervised learning. However, a theoretical understanding of self-supervised methods eludes. Addressing this, we present a generative latent variable model for self-supervised learning and show that several families of discriminative SSL, including contrastive methods, induce a comparable distribution over representations, providing a unifying theoretical framework for these methods. The proposed model also justifies connections drawn to mutual information and the use of a "projection head". Learning representations by fitting the model generatively (termed SimVAE) improves performance over discriminative and other VAE-based methods on simple image benchmarks and significantly narrows the gap between generative and discriminative representation learning in more complex settings. Importantly, as our analysis predicts, SimVAE outperforms self-supervised learning where style information is required, taking an important step toward understanding self-supervised methods and achieving task-agnostic representations. Copyright © 2024, The Authors. All rights reserved.

关键词： Self-supervised learning

RESIDUAL LIFE PREDICTION OF LITHIUM BATTERY BASED ON IMPROVED MULTI-OPTIMIZATION ALGORITHM AND SUPPORT VECTOR REGRESSION 14

学校读者我要写书评

暂无评论

RESIDUAL LIFE PREDICTION OF LITHIUM BATTERY BASED ON IMPROVE...

14th International Conference on Quality, Reliability, Risk, Maintenance, and Safety Engineering, QR2MSE 2024

作者： Tao, Yulong Wang, Jun Wang, Kai School of Computer Science and Technology Shenyang University of Chemical Technology Shenyang China Key Laboratory of Intelligent Technology for Chemical Process Industry Liaoning Province Shenyang China Key Laboratory of Networked Control Systems Chinese Academy of Sciences Shenyang China Shenyang Institute of Automation Chinese Academy of Sciences Shenyang China Robotics and Artificial Intelligence Manufacturing Innovation Institute Chinese Academy of Sciences Shenyang China

ISBN: (纸本)9781837242672

The lithium-ion battery is increasingly critical in the fields of electric vehicles and sustainable energy. Accurate prediction of the Remaining Useful Life (RUL) of lithium-ion batteries is essential to mitigate risks and minimize potential losses. This paper introduces a novel fusion algorithm, FPSOGWO, which combines Particle Swarm Optimization (PSO) and Grey Wolf Optimization (GWO) with an enhanced Circle chaos strategy and a random adaptive weighting strategy. The FPSOGWO algorithm, integrated with Support Vector Regression (FPSOGWO-SVR), was utilized to predict the RUL of health factors in lithium batteries. Compared to other optimization algorithms, FPSOGWO exhibits superior optimization speed and value, effectively overcoming issues of local optimality. Specifically, when benchmarked against the Whale Optimization Algorithm-SVR (WOA-SVR), the R-Square values of FPSOGWO-SVR showed improvements of 15.62%, 1.28%, 5.67%, and 10.65% across four different types of batteries. Experimental results demonstrate that FPSOGWO-SVR provides more accurate RUL predictions for lithium batteries, highlighting its potential for broader application in battery health management systems. © The Institution of Engineering & Technology 2024.

关键词： Support vector regression

Sign Language Recognition and Classification Model to Enhance Quality of Disabled People

学校读者我要写书评

暂无评论

computers, Materials & Continua 2022年第11期73卷 3419-3432页

作者： Fadwa Alrowais Saud S.Alotaibi Sami Dhahbi Radwa Marzouk Abdullah Mohamed Anwer Mustafa Hilal Department of Computer Sciences College of Computer and Information SciencesPrincess Nourah Bint Abdulrahman UniversityRiyadh11671Saudi Arabia Department of Information Systems College of Computing and Information SystemUmm Al-Qura UniversitySaudi Arabia Department of Computer Science College of Science&Art at MahayilKing Khalid UniversitySaudi Arabia University of Tunis EL Manar Higher Institute of ComputerResearch Team on Intelligent Systems in Imaging and Artificial Vision(Siiva)–Lab LimticAryanah2036Tunisia Department of Information Systems College of Computer and Information SciencesPrincess Nourah Bint Abdulrahman UniversityRiyadh11671Saudi Arabia Research Centre Future University in EgyptNew Cairo11745Egypt Department of Computer and Self Development Preparatory Year DeanshipPrince Sattam bin Abdulaziz UniversityAlkharjSaudi Arabia

Sign language recognition can be considered as an effective solution for disabled people to communicate with *** helps them in conveying the intended information using sign languages without any *** advancements in computer vision and image processing techniques can be leveraged to detect and classify the signs used by disabled people in an effective *** optimization algorithms can be designed in a manner such that it fine tunes the hyper parameters,used in Deep Learning(DL)models as the latter considerably impacts the classification *** this motivation,the current study designs the Optimal Deep Transfer Learning Driven Sign Language Recognition and Classification(ODTL-SLRC)model for disabled *** aim of the proposed ODTL-SLRC technique is to recognize and classify sign languages used by disabled *** proposed ODTL-SLRC technique derives EfficientNet model to generate a collection of useful feature *** addition,the hyper parameters involved in EfficientNet model are fine-tuned with the help of HGSO ***,Bidirectional Long Short Term Memory(BiLSTM)technique is employed for sign language *** proposed ODTL-SLRC technique was experimentally validated using benchmark dataset and the results were inspected under several *** comparative analysis results established the superior performance of the proposed ODTL-SLRC technique over recent approaches in terms of efficiency.

关键词： Sign language image processing computer vision disabled people deep learning parameter tuning

Machine Learning for Cataract Classification/Grading on Ophthalmic Imaging Modalities:A Survey

学校读者我要写书评

暂无评论

Machine Intelligence Research 2022年第3期19卷 184-208页

作者： Xiao-Qing Zhang Yan Hu Zun-Jie Xiao Jian-Sheng Fang Risa Higashita Jiang Liu Research Institute of Trustworthy Autonomous Systems Southern University of Science and TechnologyShenzhen 518055China Department of Computer Science and Engineering Southern University of Science and TechnologyShenzhen 518055China Tomey Corporation Nagoya 4510051Japan Cixi Institute of Biomedical Engineering Ningbo Institute of Materials Technology and EngineeringChinese Academy of SciencesNingbo 315300China Guangdong Provincial Key Laboratory of Brain-inspired Intelligent Computation Southern University of Science and TechnologyShenzhen 518055China

Cataracts are the leading cause of visual impairment and blindness *** the years,researchers have achieved significant progress in developing state-of-the-art machine learning techniques for automatic cataract classification and grading,aiming to prevent cataracts early and improve clinicians′diagnosis *** survey provides a comprehensive survey of recent advances in machine learning techniques for cataract classification/grading based on ophthalmic *** summarize existing literature from two research directions:conventional machine learning methods and deep learning *** survey also provides insights into existing works of both merits and *** addition,we discuss several challenges of automatic cataract classification/grading based on machine learning techniques and present possible solutions to these challenges for future research.

关键词： Cataract classification and grading ophthalmic image machine learning deep learning

The Kernel-SME Filter with Adaptive Kernel Widths for Association-free Multi-target Tracking

学校读者我要写书评

暂无评论

The Kernel-SME Filter with Adaptive Kernel Widths for Associ...

American Control Conference (ACC)

作者： Eugen Ernst Florian Pfaff Uwe D. Hanebeck Marcus Baum Intelligent Sensor-Actuator-Systems Laboratory (ISAS) Institute for Anthropomatics and Robotics Karlsruhe Institute of Technology (KIT) Germany Institute of Computer Science Georg- August-University Göttingen Germany

Different objectives and paradigms exist for tracking multiple targets when measurements do not contain information about the target identities (IDs). The Symmetric Measurement Equation (SME) filter can be used when one is agnostic to the labels and does not attempt to assign different IDs to the different targets. We present an extension of the Kernel-SME filter that, unlike the original variant, uses adaptive kernel widths that depend on the respective uncertainty. In our evaluation, it outperformed existing SMEbased approaches, while it is only second to a more complex global nearest neighbor tracker.

关键词：