检索结果-内蒙古大学图书馆

Cloud-Native Fog robotics: Model-Based Deployment and Evaluation of Real-Time Applications

IEEE robotics AND AUTOMATION LETTERS 2025年第1期10卷 398-405页

作者： Wen, Long Zhang, Yu Rickert, Markus Lin, Jianjie Pan, Fengjunjie Knoll, Alois Tech Univ Munich Sch Computat Informat & Technol Robot Artificial Intelligence & Real Time Syst D-80333 Munich Germany Univ Bamberg Fac Informat Syst & Appl Comp Sci Multimodal Intelligent Interact D-96047 Bamberg Germany Mercedes Benz Grp RD ASF Driver Abstract D-70372 Stuttgart Germany

As the field of robotics evolves, robots become increasingly multi-functional and complex. Currently, there is a need for solutions that enhance flexibility and computational power without compromising real-time performance. The emergence of fog computing and cloud-native approaches addresses these challenges. In this paper, we integrate a microservice-based architecture with cloud-native fog robotics to investigate its performance in managing complex robotic systems and handling real-time tasks. Additionally, we apply model-based systems engineering (MBSE) to achieve automatic configuration of the architecture and to manage resource allocation efficiently. To demonstrate the feasibility and evaluate the performance of this architecture, we conduct comprehensive evaluations using both bare-metal and cloud setups, focusing particularly on real-time and machine-learning-based tasks. The experimental results indicate that a microservice-based cloud-native fog architecture offers a more stable computational environment compared to a bare-metal one, achieving over 20% reduction in the standard deviation for complex algorithms across both CPU and GPU. It delivers improved startup times, along with a 17% (wireless) and 23% (wired) faster average message transport time. Nonetheless, it exhibits a 37% slower execution time for simple CPU tasks and 3% for simple GPU tasks, though this impact is negligible in cloud-native environments where such tasks are typically deployed on bare-metal systems.

关键词： Robots Microservice architectures Computer architecture Robot kinematics Real-time systems Resource management software hardware Modeling Edge computing hardware-software integration in robotics software architecture for robotic and automation

来源：评论

学校读者我要写书评

暂无评论

Learning Robotic Manipulation of Natural Materials With Variable Properties for Construction Tasks

引用

IEEE robotics AND AUTOMATION LETTERS 2022年第2期7卷 5749-5756页

作者： Kalousdian, Nicolas Kubail Lochnicki, Grzegorz Hartmann, Valentin N. Leder, Samuel Oguz, Ozgur S. Menges, Achim Toussaint, Marc Univ Stuttgart Inst Computat Design & Construct D-70174 Stuttgart Germany Univ Stuttgart Cluster Excellence IntCDC D-70174 Stuttgart Germany Cyber Valley D-72076 Tubingen Germany Univ Stuttgart D-70174 Stuttgart Germany Univ Stuttgart Machine Learning & Robot Lab D-70174 Stuttgart Germany TU Berlin Learning & Intelligent Syst Grp D-10623 Berlin Germany Univ Stuttgart Inst Computat Desien & Construct D-70174 Stuttgart Germany Bilkent Univ Comp Engn Dept TR-06800 Ankara Turkey

The introduction of robotics and machine learning to architectural construction is leading to more efficient construction practices. So far, robotic construction has largely been implemented on standardized materials, conducting simple, predictable, and repetitive tasks. We present a novel mobile robotic system and corresponding learning approach that takes a step towards assembly of natural materials with anisotropic mechanical properties for more sustainable architectural construction. Through experiments both in simulation and in the real world, we demonstrate a dynamically adjusted curriculum and randomization approach for the problem of learning manipulation tasks involving materials with biological variability, namely bamboo. Using our approach, robots are able to transport bamboo bundles and reach to goal-positions during the assembly of bamboo structures.

关键词： AI-enabled robotics hardware-software integration in robotics robotics and automation in construction

来源：评论

学校读者我要写书评

暂无评论

OctoMap-RT: Fast Probabilistic Volumetric Mapping Using Ray-Tracing GPUs

引用

IEEE robotics AND AUTOMATION LETTERS 2023年第9期8卷 5696-5703页

作者： Min, Heajung Han, Kyung Min Kim, Young J. J. Ewha Womans Univ Dept Comp Sci & Engn Seoul 03760 South Korea

A 3D occupancy map that is accurately modeled after real-world environments is essential for reliably performing robotic tasks. Probabilistic volumetric mapping (PVM) is a well-known environment mapping method using volumetric voxel grids that represent the probability of occupancy. The main bottleneck of current CPU-based PVM, such as OctoMap, is determining voxel grids with occupied and free states using ray-shooting. In this letter, we propose an octree-based PVM, called OctoMap-RT, using a hybrid of off-the-shelf ray-tracing GPUs and CPUs to substantially improve CPU-based PVM. OctoMap-RT employs massively parallel ray-shooting using GPUs to generate occupied and free voxel grids and to update their occupancy states in parallel, and it exploits CPUs to restructure the PVM using the updated voxels. Our experiments using various large-scale real-world benchmarking environments with dense and high-resolution sensor measurements demonstrate that OctoMap-RT builds maps up to 41.2 times faster than OctoMap and 9.3 times faster than the recent SuperRay CPU implementation. Moreover, OctoMap-RT constructs a map with 0.52% higher accuracy, in terms of the number of occupancy grids, than both OctoMap and SuperRay.

关键词： Mapping simulation and animation hardware-software integration in robotics

来源：评论

学校读者我要写书评

暂无评论

hardware-Accelerated Mars Sample Localization Via Deep Transfer Learning From Photorealistic Simulations

引用

IEEE robotics AND AUTOMATION LETTERS 2022年第4期7卷 12555-12561页

作者： Castilla-Arquillo, Raul Perez-del-Pulgar, Carlos Jesus Paz-Delgado, Gonzalo Gerdes, Levin Univ Malaga Dept Automat & Syst Engn Andalucia Tech Malaga 29070 Spain European Space Agcy Automat & Robot Sect NL-2201 Noordwijk Netherlands

The goal of the Mars Sample Return campaign is to collect soil samples from the surface of Mars and return them to Earth for further study. The samples will be acquired and stored in metal tubes by the Perseverance rover and deposited on the Martian surface. As part of this campaign, it is expected that the Sample Fetch Rover will be in charge of localizing and gathering up to 35 sample tubes over 150 Martian sols. Autonomous capabilities are critical for the success of the overall campaign and for the Sample Fetch Rover in particular. This work proposes a novel system architecture for the autonomous detection and pose estimation of the sample tubes. For the detection stage, a Deep Neural Network and transfer learning from a synthetic dataset are proposed. The dataset is created from photorealistic 3D simulations of Martian scenarios. Additionally, the sample tubes poses are estimated using Computer Vision techniques such as contour detection and line fitting on the detected area. Finally, laboratory tests of the Sample Localization procedure are performed using the ExoMars Testing Rover on a Mars-like testbed. These tests validate the proposed approach in different hardware architectures, providing promising results related to the sample detection and pose estimation.

关键词： Deep learning for visual perception hardware-software integration in robotics RGB-D perception space robotics and automation transfer learning

来源：评论

学校读者我要写书评

暂无评论

From Scene Flow to Visual Odometry Through Local and Global Regularisation in Markov Random Fields

引用

IEEE robotics AND AUTOMATION LETTERS 2022年第2期7卷 4299-4306页

作者： Scona, Raluca Matsuki, Hidenobu Davison, Andrew Imperial Coll London Dyson Robot Lab London SW7 2BU England Ocado Technol Adv Technol Div Hatfield AL10 9UL Herts England

We revisit pairwise Markov Random Field (MRF) formulations for RGB-D scene flow and leverage novel advances in processor design for real-time implementations. We consider scene flow approaches which consist of data terms enforcing intensity consistency between consecutive images, together with regularisation terms which impose smoothness over the flow field. To achieve real-time operation, previous systems leveraged GPUs and implemented regularisation only between variables corresponding to neighbouring pixels. Such systems could estimate continuously deforming flow fields but the lack of global regularisation over the whole field made them ineffective for visual odometry. We leverage the GraphCore Intelligence Processing Unit (IPU) graph processor chip, which consists of 1216 independent cores called tiles, each with 256 kB local memory. The tiles are connected to an ultrafast all-to-all communication fabric which enables efficient data transmission between the tiles in an arbitrary communication pattern. We propose a distributed formulation for dense RGB-D scene flow based on Gaussian Belief Propagation which leverages the architecture of this processor to implement both local and global regularisation. Local regularisation is enforced for pairs of flow estimates whose corresponding pixels are neighbours, while global regularisation is defined for flow estimate pairs whose corresponding pixels are far from each other on the image plane. Using both types of regularisation allows our algorithm to handle a variety of in-scene motion and makes it suitable for estimating deforming scene flow, piece-wise rigid scene flow and visual odometry within the same system.

关键词： Visual odometry Smoothing methods Three-dimensional displays Real-time systems Rigidity Image motion analysis Belief propagation hardware-software integration in robotics RGB-D perception visual tracking

来源：评论

学校读者我要写书评

暂无评论

Binarized P-Network: Deep Reinforcement Learning of Robot Control from Raw Images on FPGA

引用

IEEE robotics AND AUTOMATION LETTERS 2021年第4期6卷 8545-8552页

作者： Kadokawa, Yuki Tsurumine, Yoshihisa Maisubara, Takainitsu Nara Inst Sci & Technol Grad Sch Sci & Technol Div Informat Sci Ikoma 6300192 Japan

This letter explores a deep reinforcement learning (DRL) approach for designing image-based control for edge robots to be implemented on Field Programmable Gate Arrays (FPGAs). Although FPGAs are more power-efficient than CPUs and GPUs, a typical DRL method cannot be applied since they are composed of many Logic Blocks (LBs) for high-speed logical operations but low-speed real-number operations. To cope with this problem, we propose a novel DRL algorithm called Binarized P-Network (BPN), which learns image-input control policies using Binarized Convolutional Neural Networks (BCNNs). To alleviate the instability of reinforcement learning caused by a BCNN with low function approximation accuracy, our BPN adopts a robust value update scheme called Conservative Value Iteration, which is tolerant of function approximation errors. We confirmed the BPN's effectiveness through applications to a visual tracking task in simulation and real-robot experiments with FPGA.

关键词： Reinforcement learning embedded systems for robotic and automation hardware-software integration in robotics

来源：评论

学校读者我要写书评

暂无评论

hardware-in-the-Loop Soft Robotic Testing Framework Using an Actor-Critic Deep Reinforcement Learning Algorithm

引用

IEEE robotics AND AUTOMATION LETTERS 2023年第9期8卷 6076-6082页

作者： Marquez, Jesus Sullivan, Charles Price, Ryan M. Roberts, Robert C. Univ Texas El Paso Dept Elect & Comp Engn El Paso TX 79968 USA Univ Texas El Paso El Paso TX 79968 USA Redwire Space Albuquerque NM 87102 USA

Polymer-based soft robots are difficult to characterize due to their non-linear nature. This difficulty is compounded by multiple additional degrees of movement freedom which adds complexity to any control strategy proposed. The following work proposes and demonstrates a modular framework to test, debug and characterize soft robots using the robot operating system (ROS), to enable modeless deep reinforcement learning control strategies through hardware-in-the-loop system training. The framework is demonstrated using an actor-critic algorithm to learn a locomotion policy for a two-actuator pneu-net soft robot with integrated resistive flex sensors. The result of convergent locomotion studies was an 89.5% increase in the likelihood of reaching the end of frame design goal versus random oracle actuation vectors.

关键词： AI-enabled robotics hardware-software integration in robotics machine learning for robot control modeling control and learning for soft robots soft robot materials and design reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA

引用

IEEE robotics AND AUTOMATION LETTERS 2021年第2期6卷 2335-2342页

作者： Plancher, Brian Neuman, Sabrina M. Bourgeat, Thomas Kuindersma, Scott Devadas, Srinivas Reddi, Vijay Janapa Harvard Univ John A Paulson Sch Engn & Appl Sci Cambridge MA 02478 USA MIT Comp Sci & Artificial Intelligence Lab 77 Massachusetts Ave Cambridge MA 02139 USA Boston Dynam Waltham MA 02451 USA

Computing the gradient of rigid body dynamics is a central operation in many state-of-the-art planning and control algorithms in robotics. Parallel computing platforms such as GPUs and FPGAs can offer performance gains for algorithms with hardware-compatible computational structures. In this letter, we detail the designs of three faster than state-of-the-art implementations of the gradient of rigid body dynamics on a CPU, GPU, and FPGA. Our optimized FPGA and GPU implementations provide as much as a 3.0x end-to-end speedup over our optimized CPU implementation by refactoring the algorithm to exploit its computational features, e.g., parallelism at different granularities. We also find that the relative performance across hardware platforms depends on the number of parallel gradient evaluations required.

关键词： Computer architecture for robotics and automation hardware-software integration in robotics dynamics

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：