检索结果-内蒙古大学图书馆

Extending Maps with Semantic and Contextual Object Information for robot navigation: a Learning-Based Framework Using visual and Depth Cues

引用

JOURNAL OF INTELLIGENT & robotIC SYSTEMS 2020年第3-4期99卷 555-569页

作者： Martins, Renato Bersan, Dhiego Campos, Mario F. M. Nascimento, Erickson R. Univ Fed Minas Gerais Belo Horizonte MG Brazil Inria Sophia Antipolis France

This paper addresses the problem of building augmented metric representations of scenes with semantic information from RGB-D images. We propose a complete framework to create an enhanced map representation of the environment with object-level information to be used in several applications such as human-robot interaction, assistive robotics, visual navigation, or in manipulation tasks. Our formulation leverages a CNN-based object detector (Yolo) with a 3D model-based segmentation technique to perform instance semantic segmentation, and to localize, identify, and track different classes of objects in the scene. The tracking and positioning of semantic classes is done with a dictionary of Kalman filters in order to combine sensor measurements over time and then providing more accurate maps. The formulation is designed to identify and to disregard dynamic objects in order to obtain a medium-term invariant map representation. The proposed method was evaluated with collected and publicly available RGB-D data sequences acquired in different indoor scenes. Experimental results show the potential of the technique to produce augmented semantic maps containing several objects (notably doors). We also provide to the community a dataset composed of annotated object classes (doors, fire extinguishers, benches, water fountains) and their positioning, as well as the source code as ROS packages.

关键词： Semantic object mapping RGB-D localization and mapping Object detection visual robot navigation

来源：评论

学校读者我要写书评

暂无评论

Transfer Learning for Humanoid robot Appearance-Based Localization in a visual Map

引用

IEEE ACCESS 2021年 9卷 6868-6877页

作者： Ovalle-Magallanes, Emmanuel Aldana-Murillo, Noe G. Avina-Cervantes, Juan Gabriel Ruiz-Pinales, Jose Cepeda-Negrete, Jonathan Ledesma, Sergio Univ Guanajuato Engn Div DICIS Telemat CA & Digital Signal Proc CA Grp Campus Irapuato Salamanca Salamanca 36885 Mexico Ctr Res Math CIMAT Comp Sci Dept Guanajuato 36000 Mexico Univ Guanajuato Dept Agr Engn Div Life Sci Campus Irapuato Salamanca Irapuato 36500 Mexico

Autonomous robot visual navigation is a fundamental locomotion task based on extracting relevant features from images taken from the surrounded environment to control an independent displacement. In the navigation, the use of a known visual map helps obtain an accurate localization, but in the absence of this map, a guided or free exploration pathway must be executed to obtain the images sequence representing the visual map. This paper presents an appearance-based localization method based on a visual map and an end-to-end Convolutional Neural Network (CNN). The CNN is initialized via transfer learning (trained using the ImageNet dataset), evaluating four state-of-the-art CNN architectures: VGG16, ResNet50, InceptionV3, and Xception. A typical pipeline for transfer learning includes changing the last layer to adapt the number of neurons according to the number of custom classes. In this work, the dense layers after the convolutional and pooling layers were substituted by a Global Average Pooling (GAP) layer, which is parameter-free. Additionally, an $L_navigation$ -norm constraint was added to the GAP layer feature descriptors, restricting the features from lying on a fixed radius hypersphere. These different pre-trained configurations were analyzed and compared using two visual maps found in the CIMAT-NAO datasets consisting of 187 and 94 images, respectively. For evaluating the localization tasks, a set of 278 and 94 images were available for each visual map, respectively. The numerical results proved that by integrating the $L_navigation$ -norm constraint in the training pipeline, the appearance-based localization performance is boosted. Specifically, the pre-trained VGG16 and Xception networks achieved the best localization results, reaching a top-3 accuracy of 90.70% and 93.62% for each dataset, respectively, overcoming the referenced approaches based on hand-crafted feature extractors.

关键词： Transfer learning convolutional neural networks robot localization visual robot navigation

来源：评论

学校读者我要写书评

暂无评论

MiNERVA: Toposemantic navigation Model based on visual Information for Indoor Enviroments 10th

MiNERVA: Toposemantic Navigation Model based on Visual Infor...

引用

10th IFAC Symposium on Intelligent Autonomous Vehicles IAV

作者： Hernandez, Alejandra C. Gomez, Clara Barber, Ramon Carlos III Univ Madrid RoboticsLab Syst Engn & Automat Dept Madrid Spain

Despite the great advances in robotics made in recent years, there is not yet a universal model that allows a robot to move independently, safely and taking into account the most important elements of the surrounding environment. This paper describes the theoretical implementation of MiNERVA. MiNERVA is a model for robot navigation in human indoor environments based on visual information, conceived to enable a robot to move autonomously with full-knowledge of the environment to enhance human-robot and robot-environment integration. When robots have to share environments with humans, perception and modelling gain importance. MiNERVA manages the information of the environment, the robot internal information and the robot-environment relation in order to understand, perceive and move intelligently and autonomously. Integration of the information and management of the uncertainty are key aspects for robot navigation. Results obtained with MiNERVA are presented to validate the approach. (C) 2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.

关键词： Autonomous mobile robots Scene recognition Hierarchical models visual robot navigation Scene understanding

来源：评论

学校读者我要写书评

暂无评论

MiNERVA: Toposemantic navigation Model based on visual Information for Indoor Enviroments

引用

IFAC-PapersOnLine 2019年第8期52卷 43-48页

作者： Alejandra C. Hernandez Clara Gomez Ramon Barber System Engineering and Automation Dept. RoboticsLab Carlos III University of Madrid

关键词： Autonomous mobile robots Scene recognition Hierarchical models visual robot navigation Scene understanding

来源：评论

学校读者我要写书评

暂无评论

A Virtual Simulation of the Image Based Self-navigation of Mobile robots 4th

引用

4th Computer Science On-line Conference (CSOC)

作者： Teclaw, Mateusz Lech, Piotr Okarma, Krzysztof West Pomeranian Univ Technol Fac Elect Engn Dept Signal Proc & Multimedia Engn PL-71126 Szczecin Poland

ISBN: (纸本)9783319185033;9783319185026

The paper concerns with the problem of fully visual self-navigation of mobile robots based on the analysis of similarity of images, acquired by the cameras mounted on the robot, with some previously captured images stored in a database. In order to simplify and speed-up the extraction of the necessary data from the image database it is assumed that the rough position of the robot is known e.g. based on the GPS module or some other sensors. Due to the application of the image analysis methods, the accuracy of the self-positioning of the robot can be significantly improved leading to fully visual self-navigation of autonomous mobile robots, assuming their continuous access to the image database. In order to verify the validity of the proposed approach, the virtual simulation environment based on the Simbad 3D robot simulator has been prepared. The initial results presented in the paper, obtained for synthetic images captured by the virtual robots, confirm the usefulness of the proposed approach being a good starting point for future experiments using the real images captured by the physical mobile robot also in various lighting conditions.

关键词： machine vision visual robot navigation mobile robots

来源：评论

学校读者我要写书评

暂无评论

Construction of Approximation Spaces for Reinforcement Learning

引用

JOURNAL OF MACHINE LEARNING RESEARCH 2013年第7期14卷 2067-2118页

作者： Boehmer, Wendelin Gruenewaelder, Steffen Shen, Yun Musial, Marek Obermayer, Klaus Tech Univ Berlin Neural Informat Proc Grp D-10587 Berlin Germany UCL Ctr Computat Stat & Machine Learning London WC1E 6BT England Tech Univ Berlin Robot Grp D-10587 Berlin Germany

Linear reinforcement learning (RL) algorithms like least-squares temporal difference learning (LSTD) require basis functions that span approximation spaces of potential value functions. This article investigates methods to construct these bases from samples. We hypothesize that an ideal approximation spaces should encode diffusion distances and that slow feature analysis (SFA) constructs such spaces. To validate our hypothesis we provide theoretical statements about the LSTD value approximation error and induced metric of approximation spaces constructed by SFA and the state-of-the-art methods Krylov bases and proto-value functions (PVF). In particular, we prove that SFA minimizes the average (over all tasks in the same environment) bound on the above approximation error. Compared to other methods, SFA is very sensitive to sampling and can sometimes fail to encode the whole state space. We derive a novel importance sampling modification to compensate for this effect. Finally, the LSTD and least squares policy iteration (LSPI) performance of approximation spaces constructed by Krylov bases, PVF, SFA and PCA is compared in benchmark tasks and a visual robot navigation experiment (both in a realistic simulation and with a robot). The results support our hypothesis and suggest that (i) SFA provides subspace-invariant features for MDPs with self-adjoint transition operators, which allows strong guarantees on the approximation error, (ii) the modified SFA algorithm is best suited for LSPI in both discrete and continuous state spaces and (iii) approximation spaces encoding diffusion distances facilitate LSPI performance.

关键词： reinforcement learning diffusion distance proto value functions slow feature analysis least-squares policy iteration visual robot navigation

来源：评论

学校读者我要写书评

暂无评论

Construction of approximation spaces for reinforcement learning

The Journal of Machine Learning Research

引用

The Journal of Machine Learning Research 2013年第1期14卷

作者： Kevin Murphy Bernhard Schölkopf Wendelin Böhmer Steffen Grünewälder Yun Shen Marek Musial Klaus Obermayer Google MPI for Intelligent Systems Neural Information Processing Group Technische Universität Berlin Berlin Germany Centre for Computational Statistics and Machine Learning University College London London United Kingdom Robotics Group Technische Universität Berlin Berlin Germany

关键词： diffusion distance least-squares policy iteration proto value functions reinforcement learning slow feature analysis visual robot navigation

来源：评论

学校读者我要写书评

暂无评论

Detecting salient cues through illumination-invariant color ratios

引用

robotICS AND AUTONOMOUS SYSTEMS 2004年第2-3期48卷 111-130页

作者： Todt, E Torras, C UPC CSIC Inst Robot & Informat Ind E-08028 Barcelona Spain

This work presents a novel technique for embedding color constancy into a saliency-based system for detecting potential landmarks in outdoor environments. Since multiscale color opponencies are among the ingredients determining saliency, the idea is to make such opponencies directly invariant to illumination variations, rather than enforcing the invariance of colors themselves. The new technique is compared against the alternative approach of preprocessing the images with a color constancy procedure before entering the saliency system. The first procedure used in the experimental comparison is the well-known image conversion to chromaticity space, and the second one is based on successive lighting intensity and illuminant color normalizations. The proposed technique offers significant advantages over the preceding two ones since, at a lower computational cost, it exhibits higher stability in front of illumination variations and even of slight viewpoint changes, resulting in a better correspondence of visual saliency to potential landmark elements. (C) 2004 Elsevier B.V. All rights reserved.

关键词： visual landmarks color constancy visual saliency visual robot navigation

来源：评论

学校读者我要写书评

暂无评论

Qualitative vision for the guidance of legged robots in unstructured environments

引用

PATTERN RECOGNITION 2001年第8期34卷 1585-1599页

作者： Martínez, E Torras, C Enginyeria La Salle Dept Comunicac & Teoria Senyal Barcelona 08022 Spain UPC CSIC Inst Robot & Informat Ind Barcelona 08034 Spain

visual procedures especially tailored to the constraints and requirements of a legged robot are presented. They work for an uncalibrated camera, with pan and zoom, freely moving towards a stationary target in an unstructured environment that may contain independently moving objects. The goal is to dynamically analyse the image sequence in order to extract information about the robot motion, the target position and the environment structure. The development is based on the deformations of an active contour fitted to the target. Experimental results confirm that the proposed approach constitutes a promising alternative to the prevailing trend based on the costly computation of displacement or velocity fields. (C) 2001 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.

关键词： image sequences egomotion recovery 3D reconstruction uncalibrated camera unstructured environment contour tracking visual robot navigation

来源：评论

学校读者我要写书评

暂无评论

ROBVISION - Vision based navigation for mobile robots

引用

International Conference on Multisensor Fusion and Integration for Intelligent Systems

作者： Ponweiser, W Ayromlou, M Vincze, N Beltran, C Madsen, O Gasteratos, A Univ Vienna Inst Flexible Automat Vienna Austria

ISBN: (纸本)3000082603

This paper introduces the system, developed during the Esprit project RobVision (ROBust VIsion for Sensing in Industrial Operations and Needs), that navigates a climbing robot through a ship section for inspection and welding tasks. The basic idea is to continuously generate a robot position and orientation (pose) signal by matching visual sensing information from the environment with predetermined CAD-information. The key for robust behaviour is the integration of two diffierent vision methods, one measuring 3D junctions with a stereo head, the second tracking edge and junction features in a single image. To render tracking robust and fast, model knowledge such as feature topology, object side, and view dependent information is utilised. The pose calculation step then integrates the finding of both vision systems, detects outliers and sends the result to the robot. The real-time capability is important to reach an acceptable performance of the overall system. Presently a pose update cycle time of 120ms is achieved. Due to appearing jerks of the robot accelerometers are used for stabilization. Experiments show that our approach is feasible and reaches the positioning accuracies demanded.

关键词： CAD based vision visual robot navigation system architecture

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：