检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Ding, Jin Zhao, Jie-Chao Sun, Yong-Zhi Tan, Ping Ma, Ji-En Fang, You-Tong School of Automation and Electrical Engineering Key Institute of Robotics of Zhejiang Province Zhejiang University of Science and Technology Hangzhou310023 China School of Electrical Engineering Zhejiang University Hangzhou310027 China

Deep convolutional neural network (DCNN for short) models are vulnerable to examples with small perturbations. Adversarial training (AT for short) is a widely used approach to enhance the robustness of DCNN models by data augmentation. In AT, the DCNN models are trained with clean examples and adversarial examples (AE for short) which are generated using a specific attack method, aiming to gain ability to defend themselves when facing the unseen AEs. However, in practice, the trained DCNN models are often fooled by the AEs generated by the novel attack methods. This naturally raises a question: can a DCNN model learn certain features which are insensitive to small perturbations, and further defend itself no matter what attack methods are presented. To answer this question, this paper makes a beginning effort by proposing a shallow binary feature module (SBFM for short), which can be integrated into any popular backbone. The SBFM includes two types of layers, i.e., Sobel layer and threshold layer. In Sobel layer, there are four parallel feature maps which represent horizontal, vertical, and diagonal edge features, respectively. And in threshold layer, it turns the edge features learnt by Sobel layer to the binary features, which then are feeded into the fully connected layers for classification with the features learnt by the backbone. We integrate SBFM into VGG16 and ResNet34, respectively, and conduct experiments on multiple datasets. Experimental results demonstrate, under FGSM attack with ǫ = 8/255, the SBFM integrated models can achieve averagely 35% higher accuracy than the original ones, and in CIFAR-10 and TinyImageNet datasets, the SBFM integrated models can achieve averagely 75% classification accuracy. The work in this paper shows it is promising to enhance the robustness of DCNN models through feature learning. © 2023, CC BY-NC-SA.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

RDDS: Road Disease Detection Model Based on YOLOv5

RDDS: Road Disease Detection Model Based on YOLOv5

引用

IEEE International Conference on Cyber Technology in automation, Control, and Intelligent Systems

作者： Zhiqiang Han Jun Gong Yu Qiao School of Vehicle and Transportation Engineering Taiyuan University of Science and Technology Taiyuan China State Key Laboratory of Robotics Shenyang Institute of Automation Chinese Academy of Sciences Shenyang China

ISBN: (数字)9798331506056

ISBN: (纸本)9798331506063

The utilization of intelligent vehicles for detecting road anomalies has become a relatively common phenomenon. The existing road disease detection workload is large, the robustness is not high, and small diseases are easily ignored, which brings huge challenges to road disease detection. A road disease detection model named RDDS based on YOLOv5 is proposed. Based on the YOLOv5 algorithm, we incorporate Transformer encoder blocks into the head part to form Transformer Prediction Heads (TFE), expanding their number to four. This enhancement aims to improve the model's capability to detect small targets and road diseases of different scales. In addition, we introduce the convolutional block attention model (CBAM) to make the network pay more attention to important channels and spatial locations, thereby improving the performance and generalization ability of the model. Aiming at the problems of low resolution and insufficient data volume in current road disease data sets, we propose a new road disease detection data set RDDSC. Finally, we conducted experiments on the RDDSC data set. The experimental results showed that compared with the original YOLOv5, the RDDS parameters increased slightly, while the detection accuracy was significantly improved. Compared with existing detection algorithms, RDDS has obvious advantages in road disease detection.

关键词： YOLO Accuracy Roads Predictive models Transformers Prediction algorithms Robustness Maintenance Spatial resolution Diseases

来源：评论

学校读者我要写书评

暂无评论

A Vision-Based Target Localization Method for Robot-Assisted Sonography

A Vision-Based Target Localization Method for Robot-Assisted...

引用

International Conference on automation and Computing (ICAC)

作者： Jinyan Tian Weiyong Si Ning Wang Chenguang Yang School of Automation Science and Engineering South China University of Technology Guangzhou China Bristol Robotics Lab University of the West of England Bristol UK

This article proposes a vision-based ultrasound scanning localization method that solves the issue of human pose constraints. Specifically, by proposing an improved AlphaPose network, the neck positioning for individuals with different body types is addressed. The interference caused by other moving human subjects in the scene is resolved through an improved DeepSORT tracking algorithm. The impact of environmental noise interference is reduced through depth information processing techniques. Additionally, a robotic platform for neck ultrasound scanning is constructed. This system is capable of automatically identifying and localizing moving human necks, planning trajectories for the ultrasound probe approach, and achieving real-time interaction.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prescribed performance-based adaptive dynamic surface control for pump-controlled hydraulic systems with matched and mismatched disturbances

引用

Alexandria engineering Journal 2025年 128卷 544-555页

作者： Qian-Kun Liang Yan Cai Bing-Long Wang Jin-Chun Song Hubei Key Laboratory of Aerospace Intelligent Control Technology Hubei Sanjiang Aerospace Hongfeng Control Ltd China Space Sanjiang Group Ltd Xiaogan 432003 China State Key Laboratory of Robotics Shenyang Institute of Automation Chinese Academy of Sciences Shenyang 110169 China Key Laboratory of Marine Robotics Shenyang Liaoning Province 110169 China School of Mechanical Engineering and Automation Northeastern University Shenyang 110819 China

Uncertainties, including parametric and unstructured uncertainty, are the main obstacles for hydraulic systems to achieve high precision control. This paper investigates the asymptotic tracking problem of a pump-controlled hydraulic system subject to parametric uncertainties and matched and mismatched time-varying disturbances. First, a predefined performance function is introduced into the controller design to guarantee transient and final tracking accuracy. Then, a continuous differentiable friction model is utilized for feedforward compensation, reducing the designed conservativeness and improving the tracking performance. Moreover, robust terms suppress mismatched and matched disturbances and guarantee asymptotic stability. Adaptive dynamic surface control (ADSC) with nonlinear filters is adopted to reduce the burden of online calculation and improve control accuracy. As a result, a prescribed performance-based adaptive dynamic surface controller is proposed to achieve asymptotic tracking of a pump-controlled hydraulic system. The proposed control strategy is numerically verified and experimentally validated.

关键词： Adaptive dynamic surface control Pump-controlled hydraulic systems Asymptotic tracking Prescribed performance control Disturbances

来源：评论

学校读者我要写书评

暂无评论

Research on Cooperative Scheduling System for Mobile Robots Cluster

Research on Cooperative Scheduling System for Mobile Robots ...

引用

2022 Chinese automation Congress, CAC 2022

作者： Lili, Li Wenwei, Li Weixi, Huang South China University of Technology School of Automation Science and Engineering Guangzhou China Institute of Intelligent Manufacturing Shunde Polytechnic Foshan China South China Robotics Innovation Research Institute Robotic Laboratory Foshan China

ISBN: (纸本)9781665465335

In order to meet the requirements of robots cooperation in intelligent manufacturing, a cooperative scheduling system for mobile robots cluster was designed and built based on openTCS. The architecture and components of the cooperative scheduling system for mobile robots cluster were proposed and multi-robot collaborative SLAM was discussed. Furthermore, the scheduling system was built and tested in industrial material handling scenario. The test shows the system is compatible with various types of mobile robots and can complete material handling tasks efficiently, which has high engineering practical value. © 2022 IEEE.

关键词： Mobile robots

来源：评论

学校读者我要写书评

暂无评论

An Acceleration Inference Implementation of BEVFusion with MQBench on Xavier

An Acceleration Inference Implementation of BEVFusion with M...

引用

Chinese automation Congress (CAC)

作者： Yaqin Long Xin Zan Jianru Xue Kuizhi Mei Institute of Artificial Intelligence and Robotics Xi'an Jiaotong University Xi'an China School of Automation Science and Engineering Xi'an Jiaotong University Xi'an China

The integration of sensor technology and deep learning has made multi-sensor fusion algorithms an essential component of autonomous driving perception. Among these algorithms, BEVFusion performs well in 3D perception tasks. However, its high computing cost presents challenges for deployment and application on the hardware. In this work, we propose a way that can trade off efficiency and performance to compress and deploy it on Jetson Xavier NX. We first decouple the model into three parts and then use MQBench to quantize the Fusion and Decoder of BEV in various ways. After that, we accelerate it on the hardware with the parameters of those schemes by utilizing *** analyzing the mAP and inference time of the model in different quantization factors, we obtain the optimal schemes when using FP32 as scale, per-channel as granularity and EMAMinMax as calibration algorithm, whose accuracy only experiences a marginal drop of 0.12% while the execution speed witnesses a remarkable increase of 98.3% compared to the original model. The significant improvement makes the application of BEVFusion feasible for real-time autonomous driving perception.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Path-planning for the Human-arm-like Collaborative Robot with the Capability of Infinite Rotation

Path-planning for the Human-arm-like Collaborative Robot wit...

引用

IEEE International Conference on robotics and Biomimetics

作者： Yusheng Yang Jiajia Liu Qiaoni Yang Hang Shi Yangmin Xie School of Mechatronic Engineering and Automation Shanghai University Shanghai China Shanghai Key Laboratory of Intelligent Manufacturing and Robotics Shanghai University Shanghai China

With the advantages of high safety and scalability, collaborative robots are widely used in the fields of Human-Robot Collaboration and Interaction. However, the joint limits of the robot restrict its flexibility and workspace, especially in a cluttered environment. Inspired by the motion of the human arm, whose elbow joint and shoulder joint can rotate infinitely, a collaborative robot with the capability of infinite rotation of its first and fourth joints is constructed in this paper and named the IR-Robot. With the breakthrough of the joint limit, the corresponding dimension in the robot’s configuration space changes from a bounded dimension to an unbounded dimension. The high-dimensional torus configuration space (HTCS) is presented to describe the bounded-unbounded dimensions hybrid property of the IR-Robot’s configuration space. Additionally, an IR-RRT* algorithm is proposed to conduct path-planning in HTCS. The experimental results in simulation and the real world demonstrate the feasibility and superiority of the IR-Robot in path-following and path-planning tasks.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Multimodal Amphibious Turtle-Inspired Robot with Wheel-Leg Composite Propulsion Mechanism

A Multimodal Amphibious Turtle-Inspired Robot with Wheel-Leg...

引用

IEEE International Conference on Mechatronics and automation

作者： Yuxuan Luo Wenchuan Jia Yi Sun Shanghai Key Laboratory of Intelligent Manufacturing and Robotics School of Mechatronic Engineering and Automation Shanghai University No.99 Shangda Road Shanghai China

This paper shows a multimodal amphibious robot with wheel-leg composite propulsion mechanism (WLCPM), called AmphiTurtle-I, which achieves locomotion on land and in water with the same mechanism, with three locomotion modes: wheeled, legged and paddling swimming. The aquatic locomotion of the robot is inspired by the flattened flipper limbs of the turtle and the terrestrial locomotion is inspired by the wheel. By integrating these two types of locomotion, a WLCPM in the form of a disc with dynamically variable configuration is designed. The mechanical structure is designed and the kinematic analysis is carried out. The locomotion performance is discussed by fluid simulation and trafficability analysis. The prototype is used to carry out experiments on land and in water. Finally, AmphiTurtle-I successfully performs wheeled and legged locomotion on the ground and paddling swimming in an outdoor pond.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Robot Learning from Multiple Demonstrations Based on Generalized Gaussian Mixture Model

Robot Learning from Multiple Demonstrations Based on General...

引用

International Conference on automation and Computing (ICAC)

作者： Jiale Dong Weiyong Si Ning Wang Chenguang Yang School of Automation Science and Engineering South China University of Technology Guangzhou China Bristol Robotics Lab University of the West of England Bristol UK

The quality of skills learned by robots from demonstrations depends on the level of demonstration by the instructor, while models such as Dynamic Motion Primitives (DMP) are extremely sensitive to the quality of demonstrations because they can only learn from a single demonstration. In this paper, we propose a method of learning from multiple demonstration. First, we model multiple demonstration data by generalized Gaussian mixture models (GMMs), then generate a smooth trajectory by the Gaussian mixture regression (GMR), and finally, the generated trajectory is modeled by DMP. This method can capture features from multiple demonstrations of the same task, reducing the impact of incorrect demonstrations on skill learning. Additionally, the generalized Gaussian mixture model has strong data fitting ability, enabling robots to learn smooth and high-quality motion skills. Finally, the effectiveness of the proposed method was verified by performing a Chinese character writing task on a Baxter robot.

关键词：

来源：评论

学校读者我要写书评

暂无评论

CTSTC: An Energy-Efficient Coverage Path Planning Algorithm for Unmanned Surface Vehicles

CTSTC: An Energy-Efficient Coverage Path Planning Algorithm ...

引用

IEEE International Conference on robotics and Biomimetics

作者： Xiaowei Li Junhan Huang Yangmin Xie Dong Qu Yuxuan Zhong School of Mechatronic Engineering and Automation Shanghai University Shanghai China Shanghai Key Laboratory of Intelligent Manufacturing and Robotics Shanghai University Shanghai China

Marine environment investigation and patrol operations typically encompass a range of tasks involving Unmanned Surface Vehicles (USVs) and Coverage Path Planning (CPP). The assessment of CPP performance generally revolves around three key metrics: coverage time, extent of overlapping areas, and coverage rate. However, the consideration of energy consumption metrics is often overlooked. Especially for prolonged patrol missions or extensive coverage activities in larger regions, the minimization of energy consumption plays a pivotal role in enhancing the operational endurance of the USV while sustaining its hardware efficiency. We introduce a methodology aimed at minimizing the energy consumption associated with coverage tasks performed by USVs. Our approach draws on the principles of the Scanning Tree Coverage (STC) and the Dynamic Minimum Spanning Tree (DMST) algorithms. Adapting the STC model to the distinctive marine environment, we quantify the energy consumption resulting from ocean currents and the turns of the USV as path length. We present a method for creating a set of longest edges as a crucial part of the spanning tree, which involves dividing the map into multiple elongated regions. This approach serves to simplify the complexity of the problem. Subsequently, we employ the partial DMST algorithm to further generate tree branches. This dynamic approach aims to interconnect the longest edges based on the quantified energy consumption, ultimately striving to minimize energy consumption during coverage. By employing the method, we achieve a dual advantage: a reduction in the number of required turns, coupled with a comprehensive consideration of the influence of ocean currents. This combination empowers the USV to efficiently accomplish its tasks while consuming less energy.

关键词：

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：