检索结果-内蒙古大学图书馆

SADRL: Merging human experience with machine intelligence via supervised assisted deep reinforcement learning

NEUROCOMPUTING 2022年 467卷 300-309页

作者： Li, Xiaoshuang Wang, Xiao Zheng, Xinhu Jin, Junchen Huang, Yanhao Zhang, Jun Jason Wang, Fei-Yue Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China Univ Chinese Acad Sci Sch Artificial Intelligence Beijing 100049 Peoples R China Qingdao Acad Intelligent Ind Parallel Intelligence Res Ctr Qingdao 266109 Peoples R China Univ Minnesota Dept Elect & Comp Engn Minneapolis MN 55455 USA PCITECH PCI Intelligent Bldg2 Xincen Fourth Rd Guangzhou 510653 Peoples R China China Elect Power Res Inst State Key Lab Power Grid Safety & Energy Conserva Beijing 100192 Peoples R China Wuhan Univ Sch Elect Engn & Automat Wuhan 430072 Peoples R China

Deep Reinforcement Learning (DRL) has proven its capability to learn optimal policies in decision-making problems by directly interacting with environments. Meanwhile, supervised learning methods also show great capability of learning from data. However, how to combine DRL with supervised learning and leverage additional knowledge and data to assist the DRL agent remains difficult. This study proposes a novel Supervised Assisted Deep Reinforcement Learning (SADRL) framework integrating deep Q-learning from dynamic demonstrations with a behavioral cloning model (DQfDD-BC). Specifically, the proposed DQfDDBC method leverages historical demonstrations to pre-train a behavioral cloning model and consistently update it by learning the dynamically updated demonstrations. A supervised expert loss function is designed to compare actions generated by the DRL model with those obtained from the BC model to provide advantageous guidance for policy improvements. Experimental results in several OpenAI Gym environments show that the proposed approach accelerates the learning processes, and meanwhile, adapts to different performance levels of demonstrations. As illustrated in an ablation study, the dynamic demonstration and expert loss mechanisms using a BC model contribute to improving the learning convergence performance compared with the baseline models. We believe that SADRL provides an elegant framework and the proposed method can promote the integration of human experience and machine intelligence. (c) 2021 Elsevier B.V. All rights reserved.

关键词： Deep reinforcement learning Behavioral cloning Dynamic demonstration Double DQN

来源：评论

学校读者我要写书评

暂无评论

Learning Pre-Grasp Manipulation of Flat Objects in Cluttered Environments using Sliding Primitives

Learning Pre-Grasp Manipulation of Flat Objects in Cluttered...

引用

IEEE International Conference on Robotics and Automation (ICRA)

作者： Jiaxi Wu Haoran Wu Shanlin Zhong Quqin Sun Yinlin Li State Key Laboratory of Turbulence and Complex Systems Intelligent Biomimetic Design Lab College of Engineering Peking University Beijing China Department of Automation University of Science and Technology of China Heifei China State Key Laboratory of Management and Control for Complex System Institute of Automation Chinese Academy of Science Beijing China Science and Technology on Thermal Energy and Power Laboratory Wuhan China

Flat objects with negligible thicknesses like books and disks are challenging to be grasped by the robot because of the width limit of the robot's gripper, especially when they are in cluttered environments. Pre-grasp manipulation is conducive to rearranging objects on the table and moving the flat objects to the table edge, making them graspable. In this paper, we formulate this task as Parameterized Action Markov Decision Process, and a novel method based on deep reinforcement learning is proposed to address this problem by introducing sliding primitives as actions. A weight-sharing policy network is utilized to predict the sliding primitive's parameters for each object, and a Q-network is adopted to select the acted object among all the candidates on the table. Meanwhile, via integrating a curriculum learning scheme, our method can be scaled to cluttered environments with more objects. In both simulation and real-world experiments, our method surpasses the existing methods and achieves pre-grasp manipulation with higher task success rates and fewer action steps. Without fine-tuning, it can be generalized to novel shapes and household objects with more than 85% success rates in the real world. Videos and supplementary materials are availab.e at https://***/view/pre-grasp-sliding.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Real-time In-situ Defect Detection in 3D Printing of Continuous Fiber-Reinforced Composites

Real-time In-situ Defect Detection in 3D Printing of Continu...

引用

Digital Twins and Parallel Intelligence (DTPI), IEEE International Conference on

作者： Zhijian Yue Zhen Shen Xiangyang Dong Qihang Fang Weixing Wang Xisong Dong Gang Xiong Fei-Yue Wang School of Artificial Intelligence University of Chinese Academy of Sciences Beijing China State Key Laboratory of Multimodal Artificial Intelligence Systems Beijing Engineering Research Center of Intelligent Systems and Technology Institute of Automation Chinese Academy of Sciences Beijing China Intelligent Manufacturing Center Qingdao Academy of Intelligent Industries Qingdao China Guangdong Engineering Research Center of 3D Printing and Intelligent Manufacturing Cloud Computing Center Chinese Academy of Sciences Dongguan China State Key Laboratory for Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China

ISBN: (数字)9798350349252

ISBN: (纸本)9798350349269

With the development of continuous fiber-reinforced composites (CFRCs) 3D printing technology, timely, efficient and accurate detection of fiber path defects is essential for ensuring product quality and performance. However, challenges still need to be addressed in this field, such as inadequate training data coverage and the complexity of fiber path structures. These factors hinder the capability of traditional methods to accurately assess the status of printed products. Therefore, developing accurate and robust fiber path defect detection techniques is crucial for CFRCs 3D printing technology. In this paper, we propose a precise detection method for fiber path defects in CFRCs 3D printing using the YOLOv7 object detection model combined with a Squeeze-and-Excitation (SE) attention mechanism. By introducing the SE attention mechanism, we enhance the perception capability of the YOLOv7 model for crucial features, thereby improving the accuracy of fiber path defect detection. The experimental results show that the YOLOv7 model, when augmented with an attention mechanism, achieves an average accuracy of 93.9% in identifying fiber path defects during the printing process. Compared to models without this attention mechanism, the results show that there is 10.7% increase, with a notably enhanced learning capacity during the training phase, which substantiates the efficacy and practicality of the proposed approach.

关键词： Training Solid modeling Accuracy Attention mechanisms Training data Three-dimensional printing Real-time systems Quality assessment Optical fiber testing Defect detection

来源：评论

学校读者我要写书评

暂无评论

A PREA and OODA Computing Framework of Unmanned systems to Perform complex Tasks

IEEE Transactions on Intelligent Vehicles

引用

IEEE Transactions on intelligent Vehicles 2024年 1-18页

作者： Yang, Dongsheng Yu, Lei Zhang, Yi Zhu, Fenghua Li, Qiang Cui, Huachao Ma, Qiongmin Fan, Lili Tian, Yonglin Jinan University Guangzhou China Intelligent Science & Technology Academy Limited of CASIC Beijing China State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation Chinese Academy of Sciences Beijing China State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China 28th Research Institute of China Electronics Technology Group Corporation Nanjing China National Key Laboratory for Complex Systems Simulation National University of Defense Technology Changsha China School of Information and Electronics Beijing Institute of Technology Beijing China

A C2 computing framework for unmanned systems to perform complex tasks is proposed using methods of system of systems engineering, which is formed by combining the macro-scale command and control (C2) process mechanism model (PREA loop) and micro-scale C2 process mechanism model (OODA loop). Guided by PREA loop and OODA loop, the computing framework is divided into four steps and three kinds of transformations. The four steps are design, construction, operation monitoring, and assessment. The three transformations are tactical feedback, campaign feedback, and strategic feedback. A C2 organization model of joint landing combat force is established against the background of a typical complex task of joint landing operation, and the C2 computing framework based on the PREA &OODA is used to implement the C2 activities of the joint landing combat force. The influence of the computing framework on the performance of unmanned systems under different task conditions is verified, including task coordination load, task efficiency and sensitivity of unmanned systems response. Authors

关键词： Feedback

来源：评论

学校读者我要写书评

暂无评论

Asynchronous L1-gain Filtering of Nonlinear Positive Semi-Markov Jump systems with Time-varying Delays

Asynchronous L1-gain Filtering of Nonlinear Positive Semi-Ma...

引用

第41届中国控制会议

作者： Chao Ma Hang Fu Yidao Ji Wei Wu School of Mechanical Engineering University of Science and Technology State Key Lab of Management and Control for Complex Systems Institute of AutomationChinese Academy of Sciences

In this paper,the L-gain based filtering problem for nonlinear positive semi-Markov jump systems is investigated by proposing a novel asynchronous design *** precisely,the mode-dependent filters are designed in terms of practical observed modes instead of true system modes,such that less conservatism can be *** addition,the effect of time-varying delays is taken into account for more robustness and *** selecting suitable stochastic Lyapunov-Krasovskii functions and applying the linear programming method,sufficient conditions are established to fulfill the desired L-gain ***,the illustrative simulation is performed to verify the effectiveness of our developed control scheme.

关键词： Positive semi-Markov jump systems nonlinear semi-Markov jump systems asynchronous filtering L1-gain

来源：评论

学校读者我要写书评

暂无评论

System Dynamics-Based Analysis on Factors Influencing Comprehensive Competence of Primary and Secondary School Students

System Dynamics-Based Analysis on Factors Influencing Compre...

引用

Chinese Automation Congress (CAC)

作者： Sifeng Jing Xiwei Liu Xiaoyan Gong Hongxia Zhao The State Key Laboratory for Management and Control of Complex Systems Chinese Academy of Sciences Institute of Automation Beijing China Institute of Smart Education Systems Qingdao Academy of Intelligent Industries Qingdao China

With rapid progress of artificial intelligence (AI) , it is urgent to promote educational reform to improve comprehensive competence level of primary and secondary school students (CCLPSSS). However, educational reform is a system engineering involving multiple and intertwined factors. In this research, a system dynamics-based model of education reform is developed. Then, based on this model, a series of simulation experiments are carried out to find main factors affecting improvement of CCLPSSS. Results show that: 1) this model could simulate trends of CCLPSSS under different educational reform measures; 2) Measures including teaching mode reform, development and application of AI-based new teaching method, implementation of Double Reduction in compulsory education, implementation of deepening education evaluation reform policy, and development and application of AI-based education evaluation technology could play significant roles in promoting CCLPSSS and further strengthening cultivation of AI compound talents.

关键词： Analytical models Text analysis System dynamics Education Decision making Predictive models Market research

来源：评论

学校读者我要写书评

暂无评论

An Immersive Learning System with Multimodal Cognitive Processes Inference

An Immersive Learning System with Multimodal Cognitive Proce...

引用

Chinese Automation Congress (CAC)

作者： Siyu Bian Xiwei Liu Hongxia Zhao Xiaoyan Gong Sifeng Jing The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China Qingdao Academy of Intelligent Industries Institute of Smart Education Systems Qingdao China

Using multimodal technology to infer learners’ cognitive processes is an important means to support personalized learning. Previous cognitive process analysis is more about subjective reasoning, and less about combining physiological signals. With the rapid development of virtual reality technology, new requirements and more immersive schemes for learner’s cognitive process inferring have been proposed. Based on this, this paper proposes an immersive learning system with multimodal cognitive processes inference, which is composed of macro, mesoscopic and micro environments. The system analyzes learner’s actions, emotions and attention to infer learner’s cognitive process through more comprehensive and objective multimodal data. The system also combines eye movement, brainwaves and other data, and proposes a multimodal data acquisition and analysis scheme using HMD-based VR. This paper also provides an immersive learning cognitive processes inference scheme for individual online learning, offline classroom learning scenarios.

关键词： Learning systems Automation Cognitive processes Data acquisition Virtual reality Physiology

来源：评论

学校读者我要写书评

暂无评论

Research on intelligent methods in virtual reality games

Research on intelligent methods in virtual reality games

引用

Chinese Automation Congress (CAC)

作者： JiaWei Wang Xiwei Liu Xiaoyan Gong The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China Institute of Smart Education Systems Qingdao Academy of Intelligent Industries Qingdao China

Introduction of virtual reality technology enriches the expression of traditional games and greatly improves the sense of substitution and interactive experience of users. Realization of virtual reality is inseparable from artificial intelligence technology. In order to improve immersion and interactivity of virtual reality games, it is necessary to propose more intelligent solutions to existing technologies. This paper aims to discuss some intelligent methods of virtual reality games. Firstly, it discusses artificial intelligence means to enhance the virtual reality experience, which can improve the tracking and force / tactile interaction system of existing virtual reality games from software and hardware aspects; Then it proposes the concept of human-machine hybrid enhanced intelligence: by introducing human cognitive model to improve perception and decision-making ability of virtual reality games, so as to make interaction of virtual reality games more friendly and more real.

关键词： Solid modeling Human intelligence Force Virtual reality Games Reinforcement learning Software

来源：评论

学校读者我要写书评

暂无评论

A Multi-Modal Edge Consistency Metric Based on Regression Robustness of Truncated SVD

引用

IEEE SIGNAL PROCESSING LETTERS 2021年 28卷 1065-1069页

作者： Zhu, Mingzhu Yu, Junzhi Gao, Zhang He, Bingwei Liu, Jiantao Fuzhou Univ Sch Mech Engn & Automat Fuzhou 350108 Peoples R China Chinese Acad Sci Shenzhen Inst Adv Technol Guangdong Prov Key Lab Robot & Intelligent Syst Shenzhen 518055 Peoples R China Peking Univ State Key Lab Turbulence & Complex Syst Dept Adv Mfg & Robot BIC ESATColl Engn Beijing 100871 Peoples R China Chinese Acad Sci Inst Automat State Key Lab Management & Control Complex Syst Beijing 100190 Peoples R China

In this paper, we propose a novel edge consistency metric for multi-modal correspondence. It is based on a novel observation on image truncated SVD (singular value decomposition) termed regression robustness, which describes the fact that, a good approximation from image truncated SVD can be inherited even if the eigen-images change due to expansion and channel-dependent offsets. Compared to state-of-the-arts, multi-modal edge consistency metric can simultaneously handle multiple images with complex modality changes, including local variation, gradient reverse, intensity order change, and texture loss. Its complexity is almost linear to pixel number. Remarkable accuracies have been achieved in experiments.

关键词： Image edge detection Measurement Robustness Linear regression complexity theory Noise measurement Mutual information Regression robustness multi-modal corres-pondence edge consistency singular value decomposition

来源：评论

学校读者我要写书评

暂无评论

Collab.rative Scheduling of Computing Tasks for Edge Computing

Collaborative Scheduling of Computing Tasks for Edge Computi...

引用

2021 China Automation Congress, CAC 2021

作者： Liu, Yue Xiong, Gang Zhu, Fenghua Chen, Shichao Zhang, Liguo Liu, Xudong School of Artificial Intelligence and Automation Beijing University of Technology The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China Beijing Engineering Research Center of Intelligent Systems and Technology Institute of Automation The Cloud Computing Center Chinese Academy of Sciences Beijing China The State Key Laboratory for Management and Control of Complex Systems Institute of Automation Qingdao Academy of Intelligent Industries Chinese Academy of Sciences Beijing China Faculty of Information and Technology Beijing University of Technology Beijing China

ISBN: (纸本)9781665426473

In the face of the rapid development of the IoT, the traditional cloud computing, while providing services for mobile intelligent devices, also brings problems such as network bandwidth occupation and communication delay. Edge computing can solve these problems to some extent. But too many offloaded tasks may result in an edge server becoming overloaded and execution delay of tasks increasing, the computing resources of idle edge devices are also not fully utilized. In order to make full use of the computing resources of idle edge devices in the local network and reduce the execution delay for computing tasks, this paper takes the average delay time for all end devices as the optimization objective and proposes an end-edge cooperative scheduling (EECS) model to schedule tasks for edge computing. The scheduling algorithm used in EECS model is improved based on MCT algorithm. We use the EdgeCloudSim to verify the EECS model and scheduling algorithm. The simulation results show that the average completion time of computing tasks in EECS model, compared with only single edge server (SEE) task computing model, can be reduced by up to three times. In addition, the system load can be balanced to a certain extent in local network. © 2021 IEEE

关键词： Edge computing

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：