检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Li, Yang Tan, Wenyi Wang, Tingrui Liang, Xinkai Pan, Quan School of Automation Northwestern Polytechnical University Xi’an710072 China Laboratory of Science and Technology on Complex System Control and Intelligent Agent Cooperation Beijing China

To achieve attacks on object detection in the physical world, many existing studies focus on adversarial camouflage generation techniques, which involve applying adversarial textures onto the surface of the target. However, the neural renderers used in previous works cannot realistically simulate the physical world, and most texture rendering processes are implemented on 2D images, without fully considering the influence of the structural information of 3D models on texture overlay. To address these challenges, we introduce a novel neural rendering approach specifically tailored for adversarial camouflage within a versatile 3D rendering framework. Our method, named Flexible Physical-camouflage Attack (FPA), goes beyond traditional techniques by faithfully simulating lighting conditions, material variations of the camouflage, the angle and distance between the target and the camera, ensuring a nuanced and realistic representation of textures on a 3D target. To achieve this, we employ a generative approach that learns adversarial patterns from a diffusion model. This involves combining specially designed adversarial loss, smoothness loss, non-printable score loss, and concealment constraint loss to ensure the adversarial, printability, and concealable nature of camouflage in the physical world. Furthermore, we showcase the effectiveness of the proposed camouflage in sticker mode, demonstrating its ability to cover the target without compromising adversarial information. Through empirical and physical experiments, FPA exhibits strong performance in terms of attack success rate and transferability. Additionally, the designed sticker-mode camouflage, coupled with a concealment constraint, adapts to the environment, yielding diverse styles of texture. Our findings highlight the versatility and efficacy of the FPA approach in adversarial camouflage applications. Copyright © 2024, The Authors. All rights reserved.

关键词： Camouflage

来源：评论

学校读者我要写书评

暂无评论

Leader-follower Formation control for a Multi-missile system via Deep Reinforcement Learning

Leader-follower Formation Control for a Multi-missile System...

引用

Chinese control and Decision Conference, CCDC

作者： Jianfeng Li Keyuan Yu Shenmin Song Center for Control Theory and Guidance Technology School of Astronautics Harbin Institute of Technology Harbin China Science and Technology on Complex System Control and Intelligent Agent Cooperation Laboratory Beijing China

ISBN: (数字)9798350387780

ISBN: (纸本)9798350387797

The methods for the formation tracking control problem are mainly based on nonlinear control theories. Usually, the control effect of such methods heavily depends on an accurate control model and a lot of controller parameters need to be selected properly. To cope with these shortcomings, a novel reinforcement learning-based formation controller is proposed for a leader-follower multi-missile system. The deep deterministic policy gradient (DDPG) algorithm is implemented to generate the formation control signals. An interaction environment containing a leader and a follower is formulated, in which a model-free optimal control policy is obtained by a training process. The convergent actor-network is tested for a multi-missile system. Simulation results show that the formation control objective can be rapidly achieved in spite of the change of leader maneuvers. There is no need to change the hyperparameters of the policy for different kinds of formation scenarios, which shows great advantages over the conventional nonlinear control theory-based methods in terms of parameter selection.

关键词： Training Three-dimensional displays Target tracking Simulation Heuristic algorithms Process control Optimal control

来源：评论

学校读者我要写书评

暂无评论

Improved RRT Autonomous Exploration Method Based on Hybrid Clustering Algorithm

Improved RRT Autonomous Exploration Method Based on Hybrid C...

引用

International Conference on Autonomous Unmanned systems, ICAUS 2021

作者： Yin, Yizhen Ma, Hongjun Liang, Xinkai College of Information Science and Engineering Northeastern University Liaoning Shenyang China Science and Technology On Complex System Control and Intelligent Agent Cooperation Laboratory Beijing Electro-Mechanical Engineering Institute Beijing China

ISBN: (纸本)9789811694912

In autonomous robot exploration, it is very important that the robot can autonomously plan the exploration path of the unknown environment and create a map representation of the surrounding environment. In the robot's autonomous environment traversal method, one branch is the frontier-based detection method. This method consumes more and more computing resources as the map expands and is not suitable for large-scale environment detection. Another branch is the random exploration method based on multiple rapidly-exploring randomized trees. This method uses the edge of the tree to search for the boundary, but the selection of the next exploration point is very demanding. This paper proposes an exploration point selection strategy based on the hybrid clustering method, which speeds up the RRT exploration method. Simulation experiments show that the improved RRT exploration method proposed in this paper takes less time to explore and build maps, especially in a large-scale mapping environment, which effectively improves the completeness and efficiency of mapping. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Cluster analysis

来源：评论

学校读者我要写书评

暂无评论

Semi-autonomous Multicopter Formation control Under One Remote Pilot in GPS-Denied Environment

Semi-autonomous Multicopter Formation Control Under One Remo...

引用

International Conference on Autonomous Unmanned systems, ICAUS 2021

作者： Gao, Yan Wei, Donghui Bai, Chenggang Quan, Quan School of Automation Science and Electrical Engineering Beihang University Beijing China Science and Technology on Complex System Control and Intelligent Agent Cooperation Laboratory Beijing Electro-Mechanical Engineering Institute Beijing China

ISBN: (纸本)9789811694912

The multicopter formation is widely used in many different complex circumstances. And the semi-autonomous multicopter formation controlled with a single pilot on the ground draws people’s attention due to the great adaptability to various kinds of environments. In this paper, we focus on the problem of passing through narrow channels with semi-autonomous multicopter formation. Similar to the definition for a single multicopter, the altitude hold mode for the multicopter formation is defined first. Then, the formation decision-making and formation low-level control of the altitude hold mode are proposed one by one. Finally, a hardware-in-the-loop simulation and a real experiment are performed with four multicopters to pass through a narrow channel. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Decision making

来源：评论

学校读者我要写书评

暂无评论

Vibration-Based Fault Diagnosis for Railway Point Machines Using VMD and Multiscale Fluctuation-Based Dispersion Entropy

引用

Chinese Journal of Electronics 2024年第3期33卷 803-813页

作者： Yongkui SUN Yuan CAO Peng LI Guo XIE Tao WEN Shuai SU National Engineering Research Center of Rail Transportation Operation and Control System Beijing Jiaotong University School of Electronic and Information Engineering Beijing Jiaotong University Shaanxi Key Laboratory of Complex System Control and Intelligent Information Processing Xi'an University of Technology Frontiers Science Center for Smart High-speed Railway System Beijing Jiaotong University

As one of the most important railway signaling equipment,railway point machines undertake the major task of ensuring train operation *** fault diagnosis for railway point machines becomes a hot *** the advantage of the anti-interference characteristics of vibration signals,this paper proposes an novel intelligent fault diagnosis method for railway point machines based on vibration signals.A feature extraction method combining variational mode decomposition(VMD) and multiscale fluctuation-based dispersion entropy is developed,which is verified a more effective tool for feature ***,a two-stage feature selection method based on Fisher discrimination and ReliefF is proposed,which is validated more powerful than single feature selection ***,support vector machine is utilized for fault *** comparisons show that the proposed method performs *** diagnosis accuracies of normal-reverse and reverse-normal switching processes reach 100% and 96.57% ***,it is a try to use new means for fault diagnosis on railway point machines,which can also provide references for similar fields.

关键词： Fault diagnosis Railway point machine Vibration signal Variational mode decomposition Two-stage feature selection

来源：评论

学校读者我要写书评

暂无评论

SeqEnhance: A Lightweight Image Processing PipeLine for Low-Light Image Enhancement 7

SeqEnhance: A Lightweight Image Processing PipeLine for Low-...

引用

7th Asian Conference on Artificial Intelligence technology, ACAIT 2023

作者： Zhou, Runlin Huang, Peiliang Zhang, Dingwen Ren, Jun Anhui University AHU-IAI Ai Joint Laboratory Hefei China Institute of Artificial Intelligence Hefei Comprehensive National Science Center Hefei China Northwestern Polytechnical University School of Automation Xi'an China Science And Technology On Complex System Control And Intelligent Agent Cooperation Laboratory Beijing China

ISBN: (纸本)9798350359145

Low-light image enhancement (LLIE) is an important task in computer vision, aiming to improve the visual perception or interpretability of images captured in poorly illuminated environments. Recently, deep learning based methods have been extensively explored to address this issue. While many of these methods have achieved significant advancements in various evaluation metrics for LLIE, only a few have made progress in improving inference speed for the resulting images. As a result, achieving both high-quality enhancements and efficient inference speed remains a challenge. To tackle this challenge, we propose SeqEnhance, a fast LLIE method based on a predefined parameterized image processing pipeline. Our approach combines the inference capabilities of deep neural networks for parameter estimation and the efficient processing capabilities of image processing pipelines to generate enhanced images in an end-to-end manner. The experimental results demonstrate that the proposed method achieves competitive performance on image quality evaluation metrics such as PSNR and SSIM with a fast inference speed. © 2023 IEEE.

关键词： Image enhancement

来源：评论

学校读者我要写书评

暂无评论

Distance estimation of obstacles in monocular vision based on deep learning 8

Distance estimation of obstacles in monocular vision based o...

引用

8th Symposium on Novel Photoelectronic Detection technology and Applications

作者： Shan, Liu Ming, Li Long Hui, Wei Dong Beijing Electro-Mechanical Engineering Institute Science and Technology on Complex System Control and Intelligent Agent Cooperation Laboratory Beijing100074 China China Aerospace Science and Industry Corporation Limited Beijing100048 China

ISBN: (数字)9781510653122

ISBN: (纸本)9781510653115

Aiming at the forward-looking distance perception requirements of obstacles such as mountains, forests and trees during the ultra-low-altitude flight of subsonic aircraft, an obstacle distance estimation method based on deep learning is proposed. Firstly, the relationship between adjacent hyper-pixels is modeled by the conditional random fields, the neural network model is trained using the data set, and then the loss function of the pose parameter regression is designed, and the representation of the neural network model is input to the constructed loss function to optimize the weight of the regression model. Simulation experiments show that the proposed method can effectively perceive obstacles within 3km ahead, and the distance estimation error is not more than 20% of the relative distance, which can meet the obstacle avoidance requirements of most sudden obstacles during ultra-low-altitude flight. © 2022 SPIE

关键词： Vision

来源：评论

学校读者我要写书评

暂无评论

Research on Location Algorithm Based on SAR and Infrared Fusion 1

引用

International Conference on Autonomous Unmanned systems, ICAUS 2021

作者： Yang, Yanjiao Qu, Guangzhou Li, Bo Ma, Meng Liu, ShuaiShuai Chi, QingXi Beijing Electro-Mechanical Engineering Institute Beijing China Science and Technology on Complex System Control and Intelligent Agent Coopration Laboratory Beijing China

ISBN: (数字)9789811694929

ISBN: (纸本)9789811694912

Aiming at the problem of sins cumulative error correction, a dual-mode navigation algorithm based on infrared and SAR information fusion is studied. Radar system can provide SAR image and range information, and can measure height under certain conditions, but it cannot work in front. Infrared system can provide infrared image and optical axis direction of infrared lens. Based on the satellite positioning algorithm, this paper studies a fusion algorithm of radar SAR image and infrared. The image matching and measurement results are used to navigate to correct the accumulated errors of sins. Simulation results show the effectiveness of the algorithm. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Radar imaging

来源：评论

学校读者我要写书评

暂无评论

A Noise Filter for Dynamic Vision Sensor Based on Spatiotemporal Correlation and Hot Pixel Detection

A Noise Filter for Dynamic Vision Sensor Based on Spatiotemp...

引用

International Conference on Autonomous Unmanned systems, ICAUS 2021

作者： Xu, Ninghui Zhao, Jiajia Ren, Yuan Wang, Lihui Key Laboratory of Micro-Inertial Instrument and Advanced Navigation Technology Ministry of Education School of Instrument Science and Engineering Southeast University Nanjing210096 China Science and Technology on Complex System Control and Intelligent Agent Cooperation Laboratory Beijing100074 China

ISBN: (纸本)9789811694912

As a new type of neuromorphic sensor, the events-driven Dynamic Vision Sensor (DVS) has the advantages of low latency and wide dynamic range. The development of DVS overcomes the application limitations of traditional cameras in the field of high speed robots and autonomous driving. However, there are abundant unwanted noise events contained in the output of DVS, which severely affects the subsequent data-processing. To tackle the problem, a novel noise filter for DVS is proposed in this paper. Firstly, since the noise is composed of hot pixel events with high-frequency and random background activities, a time window is constructed to continuously detect hot pixels. Secondly, a spatiotemporal correlation-based denoising approach for each incoming event is conducted, while excluding its surrounding hot pixels, to avoid the negative effect caused by their high temporal correlation on the filter. Finally, the hot pixels are compensated with the events that are most probably to occur depending on the adjacent events, to achieve the completeness and precision of the output event stream. Experiments in different scenes demonstrate that the proposed noise filter can effectively eliminate the noise from the event stream of DVS and obviously outperforms the baseline method. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Pixels

来源：评论

学校读者我要写书评

暂无评论

Relative navigation method for unmanned aerial vehicles formation using consensus unscented Kalman filter

Relative navigation method for unmanned aerial vehicles form...

引用

International Symposium on Autonomous systems (ISAS)

作者： Zhixing Zhuang Baichun Gong Xin Ding Mingrui Hao Linxiu Chen Nanjing University of Aeronautics and Astronautics Nanjing China Science and Technology on Complex System Control and Intelligent Agent Cooperation Laboratory Beijing China

A relative navigation method in which the measurement information between aircraft only has distance or angle was proposed to achieve accurate relative navigation between aircraft formations in the global navigation satellite system (GNSS) Denied Environment. In addition, four relative measurement schemes are analyzed. An augmented state is introduced in the relative navigation system model, and the observability analysis of the navigation system is carried out relying on the Lie derivative theory, which gives the limitation of motion: the relative position and velocity of the aircraft cannot be parallel or zero, and a single aircraft cannot keep flying straight. A consensus unscented Kalman filter (CUKF) algorithm is established using the geometric characteristics in flight. The observability of the system state is improved after using the additional constraint. Finally, a standard Monte Carlo simulation system is established, and the system simulation is carried out, which shows that the Observability analysis is verified to be correct, and the algorithm demonstrates high accuracy.

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：