检索结果-内蒙古大学图书馆

第34届中国控制与决策会议

作者： Ji Peng Li Xiaoqiang Gao Wei Li Ming Faculty of Robot Science and Engineering Northeastern University

The large-scale environmental SLAM which is based on visual information,is an important technical field of mobile robot ***,a single robot is difficult to achieve the semantic SLAM tasks because of the limited sensing range and the low computing *** problem can be effectively solved by multi-robot cooperation that is using their collected visual information to model *** this paper,a vision based multi-robot cooperative semantic SLAM algorithm is ***,to achieve the distributed map splicing that is constructed by several heterogeneous robot,a cloud computing method is also proposed ***,to satisfy the demand of low bandwidth and data transmission between robots and cloud,a high density information representation and a transmission strategy against spatiotemporal delay are *** this basis,the LEDNET model is used to obtain the semantic information of environmental feature ***,the centroid of map points is obtained to calculate the position information of objects in the environment,the attitude information of objects in the environment is determined by principal component analysis,and a multi map stitching method based on key frame is established to realize the collaborative slam of multi *** the map mosaic test of three robots on KITTI data set,the multi-robot cooperative SLAM algorithm proposed in this paper can quickly and accurately construct the semantic map of large-scale environment.

关键词： Visual SLAM Multiple robots Cloud Computing Semantic Segmentation Map Splicing

来源：评论

学校读者我要写书评

暂无评论

Monocular Dynamic Object Detection with multi-modal fusion

Monocular Dynamic Object Detection with multi-modal fusion

引用

Advanced Information, Mechanical engineering, robotics and Automation (AIMERA), IEEE International Conference on

作者： Li Li Yingcai Wan JiShen Peng Faculty of Electrical and Control Liaoning Technical University Hu Ludao China Faculty of Robot Science and Engineering Northeastern University Shen Yang China

ISBN: (数字)9798350343335

ISBN: (纸本)9798350343342

In the field of 3D scene understanding, object motion estimation holds great significance in handling dynamic scenes. To address this, This paper presents a novel method for rigid object motion segmentation from video, integrating multiple data modalities including imagery, optical flow, depth, scene flow, and camera ego-motion. Utilizing a motion filtering module, effectively removes background motion due to camera dynamics, while a motion extraction module accurately segments moving objects. The approach computes the objects’ motion characteristics such as speed, direction, and position using a sophisticated fusion of 3D motion data. Validated on the KITTI dataset, our method demonstrates superior accuracy and robustness in dynamic environments.

关键词： Computer vision Image segmentation Image motion analysis Three-dimensional displays Filtering Motion segmentation Dynamics Object detection Cameras Optical flow

来源：评论

学校读者我要写书评

暂无评论

Temporal Enhance and Spatial Gated Network for Group Activity Recognition

Temporal Enhance and Spatial Gated Network for Group Activit...

引用

第34届中国控制与决策会议

作者： Tiansheng Sun Jianning Chi Chengdong Wu Faculty of Robot Science and Engineering Northeastern University

The main goal of Group Activity Recognition(GAR) is to analyze the group behavior in a multi-actors *** is a challenging task that integrates group activity features from individual actors who frequently interact with each ***,existing state-of-the-art methods do not consider temporal information and take full advantage of spatial ***,existing methods utilize max-pooling operation to generate group activity features,which introduces noise into group activity features and leads to bad *** tackle these problems,a temporal enhance and spatial gated network is *** fully utilize the rich temporal features and spatial features,two branches are set up to extract temporal features and actor interactive relations,***,a feature aggregation module is used to generate group activity features,which can decrease the noise introduced by the max-pooling *** conduct a series of experiments on two stand *** results on these benchmarks demonstrate that the proposed approach achieves state-of-the-art *** with previous work,the recognition accuracy increases by 0.5%～1.0%.

关键词： Group Activity Recognition Convolutional Neural Network Interactive Relations

来源：评论

学校读者我要写书评

暂无评论

robot Grasping Detection in Object Overlapping Scenes Based on Multi-Stage ROI Extraction

Robot Grasping Detection in Object Overlapping Scenes Based ...

引用

第34届中国控制与决策会议

作者： Jintao Xia Jianning Chi Chengdong Wu Fengyu Zhao Faculty of Robot Science and Engineering Northeastern University

In multi-object stacking scenes,it is difficult for robots to detect and grasp *** propose a new robot grasp detection algorithm Multi-Stage ROI Grasp Detection(MSROI-GD).MSROI-GD uses multi-stage extracted ROI features to detect objects and *** algorithm can effectively filter and utilize *** results show that the improved MSROI-GD improves the accuracy of the original ROI-GD algorithm by 4.3%,and exceeds the current state-of-the-art algorithm by 0.5% in the overlapping scenes of objects in the VMRD *** the same time,using our grasp detection algorithm on the Cornell grasp dataset still has good *** experiments show that MSROI-GD can help robots grasp object in multi-object scenes with a success rate of 85%.

关键词： Grasp Detection Multi-object Stacking Scenes Multi-stage Extracted ROI

来源：评论

学校读者我要写书评

暂无评论

Synchronization Control of Delayed Inertial Neural Networks with Semi-Markovian Jumping

Synchronization Control of Delayed Inertial Neural Networks ...

引用

第42届中国控制会议

作者： Junyi Cui Huixia Cui Yingxin Zhang Junyi Wang Min Lei College of Information Science and Engineering Northeastern University Faculty of Robot Science and Engineering Northeastern University NO.710 R&D Institute CSIC

This article deals with the synchronization issue of the time-varying delayed inertial neural networks(INNs) with semi-Markovian jumping(SMJ).The synchronization criteria of the system can be obtained by using the Lyapunov-Krasovskii functional(LKF) and a general free-matrix-based integral *** addition,the mode-dependent control gain matrices are obtained based on synchronization *** last,an example is proposed to show the validity of the theoretical results.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Neural network equivalent model for highly efficient massive data classification

引用

science China(Information sciences) 2023年第3期66卷 287-288页

作者： Siquan YU Zhi HAN Yandong TANG Chengdong WU School of Information Science and Engineering Northeastern University Institutes for Robotics and Intelligent Manufacturing Chinese Academy of Sciences State Key Laboratory of Robotics Shenyang Institute of Automation Chinese Academy of Sciences Faculty of Robot Science and Engineering Northeastern University

Dear editor, Recently, neural networks(NNs) have been successfully applied to massive data classification tasks [1, 2]. However,there is an inconsistent problem in neural network structure and convergence [3]. On one hand, under the overparameterize assumption, it requires to design a complex network to guarantee the convergence [4].

关键词：

来源：评论

学校读者我要写书评

暂无评论

EvGGS: a collaborative learning framework for event-based generalizable Gaussian splatting 24

EvGGS: a collaborative learning framework for event-based ge...

引用

Proceedings of the 41st International Conference on Machine Learning

作者： Jiaxu Wang Junhao He Ziyi Zhang Mingyuan Sun Jingkai Sun Renjing Xu Function Hub Hong Kong University of Science and Technology (GZ) Guangzhou China Faculty of Robot Science and Engineering Northeastern University Shenyang China

Event cameras offer promising advantages such as high dynamic range and low latency, making them well-suited for challenging lighting conditions and fast-moving scenarios. However, reconstructing 3D scenes from raw event streams is difficult because event data is sparse and does not carry absolute color information. To release its potential in 3D reconstruction, we propose the first eventbased generalizable 3D reconstruction framework, called EvGGS, which reconstructs scenes as 3D Gaussians from only event input in a feedforward manner and can generalize to unseen cases without any retraining. This framework includes a depth estimation module, an intensity reconstruction module, and a Gaussian regression module. These submodules connect in a cascading manner, and we collaboratively train them with a designed joint loss to make them mutually promote. To facilitate related studies, we build a novel event-based 3D dataset with various material objects and calibrated labels of grayscale images, depth maps, camera poses, and silhouettes. Experiments show models that have jointly trained significantly outperform those trained individually. Our approach performs better than all baselines in reconstruction quality, and depth/intensity predictions with satisfactory rendering speed.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Deep Fourier-embedded Network for RGB and Thermal Salient Object Detection

arXiv

引用

arXiv 2024年

作者： Lyu, Pengfei Yeung, Pak-Hei Yu, Xiaosheng Wu, Chengdong Rajapakse, Jagath C. Faculty of Robot Science and Engineering Northeastern University Shenyang110169 China College of Computing and Data Science Nanyang Technological University 639798 Singapore

The rapid development of deep learning has significantly improved salient object detection (SOD) combining both RGB and thermal (RGB-T) images. However, existing deep learning-based RGB-T SOD models suffer from two major limitations. First, Transformer-based models with quadratic complexity are computationally expensive and memory-intensive, limiting their application in high-resolution bi-modal feature fusion. Second, even when these models converge to an optimal solution, there remains a frequency gap between the prediction and ground-truth. To overcome these limitations, we propose a purely Fourier transform-based model, namely Deep Fourier-Embedded Network (DFENet), for accurate RGB-T SOD. To address the computational complexity when dealing with high-resolution images, we leverage the efficiency of fast Fourier transform with linear complexity to design three key components: (1) the Modal-coordinated Perception Attention, which fuses RGB and thermal modalities with enhanced multi-dimensional representation;(2) the Frequency-decomposed Edge-aware Block, which clarifies object edges by deeply decomposing and enhancing frequency components of low-level features;and (3) the Fourier Residual Channel Attention Block, which prioritizes high-frequency information while aligning channel-wise global relationships. To mitigate the frequency gap, we propose Co-focus Frequency Loss, which dynamically weights hard frequencies during edge frequency reconstruction by cross-referencing bi-modal edge information in the Fourier domain. Extensive experiments on four RGB-T SOD benchmark datasets demonstrate that DFENet outperforms fifteen existing state-of-the-art RGB-T SOD models. Comprehensive ablation studies further validate the value and effectiveness of our newly proposed components. The code is available at https://***/JoshuaLPF/DFENet. Copyright © 2024, The Authors. All rights reserved.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Efficient Fourier Filtering Network with Contrastive Learning for UAV-based Unaligned Bi-modal Salient Object Detection

arXiv

引用

arXiv 2024年

作者： Lyu, Pengfei Yeung, Pak-Hei Cheng, Xiufei Yu, Xiaosheng Wu, Chengdong Rajapakse, Jagath C. The Faculty of Robot Science and Engineering Northeastern University Shenyang110169 China The College of Computing and Data Science Nanyang Technological University 639798 Singapore

Unmanned aerial vehicle (UAV)-based bi-modal salient object detection (BSOD) aims to segment salient objects in a scene utilizing complementary cues in unaligned RGB and thermal image pairs. However, the high computational expense of existing UAV-based BSOD models limits their applicability to real-world UAV devices. To address this problem, we propose an efficient Fourier filter network with contrastive learning that achieves both real-time and accurate performance. Specifically, we first design a semantic contrastive alignment loss to align the two modalities at the semantic level, which facilitates mutual refinement in a parameter-free way. Second, inspired by the fast Fourier transform that obtains global relevance in linear complexity, we propose synchronized alignment fusion, which aligns and fuses bi-modal features in the channel and spatial dimensions by a hierarchical filtering mechanism. Our proposed model, AlignSal, reduces the number of parameters by 70.0%, decreases the floating point operations by 49.4%, and increases the inference speed by 152.5% compared to the cutting-edge BSOD model (i.e., MROS). Extensive experiments on the UAV RGB-T 2400 and three weakly aligned datasets demonstrate that AlignSal achieves both real-time inference speed and better performance and generalizability compared to sixteen state-of-the-art BSOD models across most evaluation metrics. In addition, our ablation studies further verify AlignSal's potential in boosting the performance of existing aligned BSOD models on UAV-based unaligned data. The code is available at: https://***/JoshuaLPF/AlignSal. Copyright © 2024, The Authors. All rights reserved.

关键词： Unmanned aerial vehicles (UAV)

来源：评论

学校读者我要写书评

暂无评论

Auxiliary Linear Extended State Observer for Position Tracking Control of PMSM Based on Sliding Mode Control

Auxiliary Linear Extended State Observer for Position Tracki...

引用

Asian Control Conference

作者： Yifei Yao Yaoming Zhuang Chengdong Wu Fengkui Cao Faculty of Information Science and Engineering Northeastern University Shenyang China Faculty of Robot Science and Engineering Northeastern University Shenyang China State Key Laboratory of Robotics Shenyang Institute of Automation Chinese Academy of Sciences Shenyang China

ISBN: (数字)9789887581598

ISBN: (纸本)9798331540845

An auxiliary linear extended state observer and a position-velocity integral sliding mode controller based on improved reaching law are proposed to address position-tracking control issues in servo systems. Servo systems are always affected by internal or external disturbances, making it hard to meet the requirements of precise and fast positioning. Firstly, sliding mode control due to its merit of insensitivity to parameter variations and fast response is introduced. Secondly, an auxiliary linear extended state observer is designed for the estimation of the time-varying uncertainty, which makes up for the disadvantage of only robust to matched disturbances in sliding mode control. Next, to address the chattering issue in the sliding mode control method based on the exponential reaching law, the switching function is substituted with the saturation function, while introducing a power term of the sliding surface magnitude to track the input reference faster, resulting in an improved reaching law. Finally, the proposed approach is compared with other methods. Numer-ous simulation results demonstrate that the method we proposed improves positioning accuracy, tracks the input reference faster, reduces chattering, and has a certain robustness.

关键词： Accuracy Uncertainty Simulation Position control Switches Permanent magnet motors Robustness

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：