检索结果-内蒙古大学图书馆

Unsupervised Learning of Depth and Pose Based on Monocular Camera and Inertial Measurement Unit (IMU)

学校读者我要写书评

暂无评论

Unsupervised Learning of Depth and Pose Based on Monocular C...

IEEE International Conference on Robotics and automation (ICRA)

作者： Yanbo Wang Hanwen Yang Jianwei Cai Guangming Wang Jingchuan Wang Yi Huang University of Michigan-Shanghai Jiao Tong University Joint Institute Shanghai Jiao Tong University Shanghai China School of Electronic Information and Electrical Engineering Shanghai Jiao Tong University Shanghai China Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai China Department of Automation Shanghai Jiao Tong University Shanghai China Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai Weitong Vision Technology Co. Ltd.

The main content of the research in this paper is the estimation of depth and pose based on monocular vision and Inertial Measurement Unit (IMU). The usual depth estimation network and pose estimation network require depth ground truth or pose ground truth as a supervised signal for training, while the depth ground truth and pose ground truth are hard to obtain, and monocular vision based depth estimation cannot predict absolute depth. In this paper, with the help of IMU, which is inexpensive and widely used, we can obtain angular velocity and acceleration information. Two new supervision signals are proposed and the calculation expressions are given. Among them, the model trained with acceleration constraint shows a good ability to estimate the absolute depth during the test. It can be considered that the model can estimate the absolute depth. We also derive the method of estimating the scale factor during the test from the acceleration constraint, and also achieve good results as the acceleration constraint does. In addition, this paper also studies the method of using IMU information as pose network input and as selecting conditions. Moreover, it analyzes and discusses the experimental results. At the same time, we also evaluate the effect of the pose estimation of the relevant models. This article starts by reviewing the achievements and deficiencies of the work in this field, combines the use of IMU, puts forward three new methods such as a new loss function, and conducts a test analysis and discussion of relevant indicators on the KITTI data set.

关键词：

LinK3D: Linear keypoints Representation for 3D LiDAR Point Cloud

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Cui, Yunge Zhang, Yinlong Dong, Jiahua Sun, Haibo Chen, Xieyuanli Zhu, Feng The Key Laboratory of Opto-Electronic Information Processing Shenyang Institute of Automation Chinese Academy of Sciences Shenyang China The Key Laboratory of Networked Control Systems Shenyang Institute of Automation Chinese Academy of Sciences Shenyang China The State Key Laboratory of Robotics Shenyang Institute of Automation Chinese Academy of Science Shenyang China The Institutes for Robotics and Intelligent Manufacturing Chinese Academy of Sciences Shenyang110016 China The University of Chinese Academy of Sciences Beijing101408 China The Shanghai Institute of Microsystem and Information Technology Chinese Academy of Sciences Shanghai China The College of Intelligence Science and Technology National University of Defense Technology Changsha China

Feature extraction and matching are the basic parts of many robotic vision tasks, such as 2D or 3D object detection, recognition, and registration. As is known, 2D feature extraction and matching have already achieved great success. Unfortunately, in the field of 3D, the current methods may fail to support the extensive application of 3D LiDAR sensors in robotic vision tasks due to their poor descriptiveness and inefficiency. To address this limitation, we propose a novel 3D feature representation method: Linear keypoints representation for 3D LiDAR point cloud, called LinK3D. The novelty of LinK3D lies in that it fully considers the characteristics (such as the sparsity and complexity) of LiDAR point clouds and represents the keypoint with its robust neighbor keypoints, which provide strong constraints in the description of the keypoint. The proposed LinK3D has been evaluated on three public datasets, and the experimental results show that our method achieves great matching performance. More importantly, LinK3D also shows excellent real-time performance, faster than the sensor frame rate at 10 Hz of a typical rotating LiDAR sensor. LinK3D only takes an average of 30 milliseconds to extract features from the point cloud collected by a 64-beam LiDAR and takes merely about 20 milliseconds to match two LiDAR scans when executed on a computer with an Intel Core i7 processor. Moreover, our method can be extended to LiDAR odometry task, and shows good scalability. We release the implementation of our method at https://***/YungeCui/LinK3D. Copyright © 2022, The Authors. All rights reserved.

关键词： Object detection

FireMatch: A Semi-Supervised Video Fire Detection Network Based on Consistency and Distribution Alignment

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Lin, Qinghua Li, Zuoyong Zeng, Kun Fan, Haoyi Li, Wei Zhou, Xiaoguang School of Computer Science and Mathematics Fujian University of Technology Fuzhou350118 China Fujian Provincial Key Laboratory of Information Processing and Intelligent Control College of Computer and Control Engineering Minjiang University Fuzhou350121 China School of Computer and Artificial Intelligence Zhengzhou University Zhengzhou450001 China School of Automation Beijing University of Post and Telecommunications Beijing100876 China School of Economics and Management Minjiang University Fuzhou350121 China

Deep learning techniques have greatly enhanced the performance of fire detection in videos. However, video-based fire detection models heavily rely on labeled data, and the process of data labeling is particularly costly and time-consuming, especially when dealing with videos. Considering the limited quantity of labeled video data, we propose a semi-supervised fire detection model called FireMatch, which is based on consistency regularization and adversarial distribution alignment. Specifically, we first combine consistency regularization with pseudo-label. For unlabeled data, we design video data augmentation to obtain corresponding weakly augmented and strongly augmented samples. The proposed model predicts weakly augmented samples and retains pseudo-label above a threshold, while training on strongly augmented samples to predict these pseudo-labels for learning more robust feature representations. Secondly, we generate video cross-set augmented samples by adversarial distribution alignment to expand the training data and alleviate the decline in classification performance caused by insufficient labeled data. Finally, we introduce a fairness loss to help the model produce diverse predictions for input samples, thereby addressing the issue of high confidence with the non-fire class in fire classification scenarios. The FireMatch achieved an accuracy of 76.92% and 91.81% on two real-world fire datasets, respectively. The experimental results demonstrate that the proposed method outperforms the current state-of-the-art semi-supervised classification methods. Copyright © 2023, The Authors. All rights reserved.

关键词： Fires

Adaptive control of the Structural Balance for a Class of Complex Dynamical Networks

学校读者我要写书评

暂无评论

Journal of Systems Science & Complexity 2020年第3期33卷 725-742页

作者： GAO Zilin WANG Yinhe PENG Yi LIU Lizhi CHEN Haoguang Key Laboratory of Intelligent Information Processing and Control of Chongqing Municipal Institutions of Higher Education Chongqing Three Gorges UniversityChongqing 404100China School of Computer Science and Engineering Chongqing Three Gorges UniversityChongqing 404100China School of Automation Guangdong University of TechnologyGuangzhou 510006China School of Computer Science Zhongshan Institute of Electronic Science and Technology University of ChinaZhongshan 528402China

In social networks,the structural balance is a state of a group of individuals(nodes)with established mutual relationships(connection relationships)between *** is easy to see that a social network can be described by a complex dynamical network model composed of the nodes subsystem(NS)and the connection relationships subsystem(CS),where the two subsystems are usually coupled with each *** implies that the dynamic changes of nodes'states may cause the structural balance in ***,few papers have discussed the relationship between the structural balance and the specific dynamic changes of the nodes'*** paper proposes a model of complex dynamical networks,and mainly focuses on the dynamic changes of states in NS which can lead to the structural balance in *** is proved that if each state in NS is doing a specific dynamic motion via the controller with the parameter adaptive law,then the CS can track a given structural balance matrix via the effective coupling and the structural balance can be *** a result can be regarded as an explanation of the relationship between the structural balance and the specific dynamic changes of the nodes'***,the simulations verify the effectiveness of the proposed method.

关键词： Complex dynamical networks connection relationships effective coupling structural balance

Structural Robustness of Complex Networks: A Survey of A Posteriori Measures

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Lou, Yang Wang, Lin Chen, Guanrong The Graduate School of Information Science and Technology Osaka University Osaka Suita565-0871 Japan The Department of Computer Science National Yang Ming Chiao Tung University Hsinchu300 Taiwan The Department of Automation Shanghai Jiao Tong University Shanghai200240 China The Key Laboratory of System Control and Information Processing Ministry of Education Shanghai200240 China The Department of Electrical Engineering City University of Hong Kong Hong Kong

Network robustness is critical for various industrial and social networks against malicious attacks, which has various meanings in different research contexts and here it refers to the ability of a network to sustain its functionality when a fraction of the network fail to work due to attacks. The rapid development of complex networks research indicates special interest and great concern about the network robustness, which is essential for further analyzing and optimizing network structures towards engineering applications. This comprehensive survey distills the important findings and developments of network robustness research, focusing on the a posteriori structural robustness measures for single-layer static networks. Specifically, the a posteriori robustness measures are reviewed from four perspectives: 1) network functionality, including connectivity, controllability and communication ability, as well as their extensions;2) malicious attacks, including conventional and computation-based attack strategies;3) robustness estimation methods using either analytical approximation or machine learning-based prediction;4) network robustness optimization. Based on the existing measures, a practical threshold of network destruction is introduced, with the suggestion that network robustness should be measured only before reaching the threshold of destruction. Then, a posteriori and a priori measures are compared experimentally, revealing the advantages of the a posteriori measures. Finally, prospective research directions with respect to a posteriori robustness measures are recommended. © 2023, CC BY.

关键词： Complex networks

An advanced actor-critic algorithm for training video game ai 1st

学校读者我要写书评

暂无评论

An advanced actor-critic algorithm for training video game a...

1st International Conference on Neural Computing for Advanced Applications, NCAA 2020

作者： Zha, Zhong Yi Tang, Xue Song Wang, Bo Key Laboratory of Ministry of Education for Image Processing and Intelligent Control Artificial Intelligence and Automation School Huazhong University of Science and Technology Wuhan430074 China College of Information and Science Donghua University Shanghai China

ISBN: (纸本)9789811576690

This paper presents an improved deep reinforcement learning (DRL) algorithm, namely Advanced Actor Critic (AAC), which is based on Actor Critic (AC) algorithm, to the video game Artificial intelligence (AI) training. The advantage distribution estimator, Normal Constraint (NC) function and exploration based on confidence are introduced to improve the AC algorithm, in order to achieve accurate value estimate, select continuous spatial action correctly, and explorate effectively. The aim is to improve the performance of conventional AC algorithm in a complex environment. A case study of video game StarCraft II mini-game AI training is employed. The results verify that the improved algorithm effectively improves the performance in terms of the convergence rate, maximum reward, average reward in every 100 episode, and time to reach a specific reward, etc. The analysis on how these modifications improve the performance is also given through interpretation of the feature layers in the mini-games. © Springer Nature Singapore Pte Ltd. 2020.

关键词： Reinforcement learning

Multi-level Coordinated Energy Management for Energy Hub in Hybrid Markets with Distributionally Robust Scheduling

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Cao, Jiaxin Yang, Bo Zhu, Shanying Chung, Chi Yung Guan, Xinping Department of Automation Shanghai Jiao Tong University Shanghai200240 China Key Laboratory of System Control and Information Processing Ministry of Education China Shanghai Engineering Research Center of Industrial Intelligent Control and Management Shanghai200240 China Department of Electrical and Computer Engineering University of Saskatchewan SaskatoonSKS7N 5A9 Canada

Maintaining energy balance and economical operation is significant for multi-energy systems such as the energy hub (EH). However, it is usually challenged by the frequently changing and unpredictable uncertain parameters at different timescales. Under this scope, this paper investigates the coordinated energy management problem for day-ahead and intra-day conditions considering uncertainties of source-load and market prices concurrently. Note that the precise knowledge of distributions about uncertainties may be unaccessible before the decision-making in day-ahead phase. A two-stage chance-constrained model based on distributionally robust approach with ambiguous moment information is proposed to immunize scheduling strategies against the worst-case probability distributions. The first stage is dedicated to obtaining more energy arbitrage and operation flexibility by optimizing bidding strategies in day-ahead power, natural gas and carbon trading markets. The second stage focuses on the optimization of the worst-case expected operation cost. It provides a robust energy equipment and load scheduling strategy for the reference of subsequent intra-day arrangements. With respect to different variations of electrical and thermal components, an intra-day two-timescale coordination is implemented step by step. The energy scheduling is re-dispatched circularly to minimize the operation and penalty costs. The possible energy imbalance is also compensated by this way. As the energy management program is nonlinear, chance-constrained and multi-stage, some linearization and dual transformation techniques are designed to enhance tractability of the program. Experimental results show that the developed multi-level framework results in a carbon emission decrease of 37%, and reduces energy cost averagely 3% compared with corresponding contrasting cases. The obtained strategy validates a good tradeoff between robustness and optimality. Copyright © 2022, The Authors. All rights reser

关键词： Power markets

Hierarchical Learning for IRS-Assisted MEC Systems with Rate-Splitting Multiple Access

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wu, Yinyu Zhang, Xuhui Ren, Jinke Shen, Yanyan Yang, Bo Wang, Shuqiang Guan, Xinping Niyato, Dusit The Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Guangdong518055 China The University of Chinese Academy of Sciences Beijing100049 China The Shenzhen Future Network of Intelligence Institute The School of Science and Engineering The Guangdong Provincial Key Laboratory of Future Networks of Intelligence The Chinese University of Hong Kong Guangdong Shenzhen518172 China Shenzhen University of Advanced Technology Guangdong518055 China The Department of Automation The Key Laboratory of System Control and Information Processing Ministry of Education Shanghai Jiao Tong University Shanghai200240 China The College of Computing and Data Science Nanyang Technological University Singapore639798 Singapore

Intelligent reflecting surface (IRS)-assisted mobile edge computing (MEC) systems have shown notable improvements in efficiency, such as reduced latency, higher data rates, and better energy efficiency. However, the resource competition among users will lead to uneven allocation, increased latency, and lower throughput. Fortunately, the rate-splitting multiple access (RSMA) technique has emerged as a promising solution for managing interference and optimizing resource allocation in MEC systems. This paper studies an IRS-assisted MEC system with RSMA, aiming to jointly optimize the passive beamforming of the IRS, the active beamforming of the base station, the task offloading allocation, the transmit power of users, the ratios of public and private information allocation, and the decoding order of the RSMA to minimize the average delay from a novel uplink transmission perspective. Since the formulated problem is non-convex and the optimization variables are highly coupled, we propose a hierarchical deep reinforcement learning-based algorithm to optimize both continuous and discrete variables of the problem. Additionally, to better extract channel features, we design a novel network architecture within the policy and evaluation networks of the proposed algorithm, combining convolutional neural networks and densely connected convolutional network for feature extraction. Simulation results indicate that the proposed algorithm not only exhibits excellent convergence performance but also outperforms various benchmarks. Copyright © 2024, The Authors. All rights reserved.

关键词： Deep reinforcement learning

Memristor-based multilayer neural network with edge learning and weight update

学校读者我要写书评

暂无评论

Neurocomputing 2025年 647卷

作者： Ningye Jiang Jupeng Xie Mingxuan Jiang Haoen Huang Zhigang Zeng School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan 430074 China Key Laboratory of Image Information Processing and Intelligent Control Ministry of Education of China Wuhan 430074 China

Memristor crossbar array (MCA) is a promising computing-in-memory architecture for accelerating vector-matrix multiplication (VMM). However, existing memristor-based circuits primarily perform forward computations without local learning capabilities. To address the issue, a memristor-based multilayer neural network (MNN) with edge learning is proposed, enabling error backpropagation and one-step weight update. In particular, the overall circuit design of a three-layer MNN is outlined, comprising three operational stages: forward computation, backpropagation, and weight update. Subsequently, the schematics of four adopted circuit blocks are detailed. By integrating the memory and weight update blocks with MCA, the proposed three-layer memristor-based MNN exhibits scalability and universality, allowing it to be extended to the deep neural network (DNN) architecture. Circuit simulations validate the effectiveness of the proposed scheme, while offline-online co-simulations demonstrate its strength in compensating for variations and non-ideal characteristics of physical devices. Comparative results further highlight the advantages of the proposed scheme in realizing full-analog edge learning circuits and performing error backpropagation processes efficiently.

关键词： Analog circuit design Computing-in-memory Edge learning Memristor crossbar array (MCA) Multilayer neural network