检索结果-内蒙古大学图书馆

Zero-Shot Object Goal Visual Navigation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhao, Qianfan Zhang, Lu He, Bin Qiao, Hong Liu, Zhiyong The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China The School of Artificial Intelligence University of Chinese Academy of Sciences Beijing100190 China The Nanjing Artificial Intelligence Research of IA Jiangning District Jiangsu Nanjing211100 China The College of Electronic and Information Engineering Tongji University China

Object goal visual navigation is a challenging task that aims to guide a robot to find the target object based on its visual observation, and the target is limited to the classes predefined in the training stage. However, in real households, there may exist numerous target classes that the robot needs to deal with, and it is hard for all of these classes to be contained in the training stage. To address this challenge, we study the zero-shot object goal visual navigation task, which aims at guiding robots to find targets belonging to novel classes without any training samples. To this end, we also propose a novel zero-shot object navigation framework called semantic similarity network (SSNet). Our framework use the detection results and the cosine similarity between semantic word embeddings as input. Such type of input data has a weak correlation with classes and thus our framework has the ability to generalize the policy to novel classes. Extensive experiments on the AI2-THOR platform show that our model outperforms the baseline models in the zero-shot object navigation task, which proves the generalization ability of our model. Our code is available at: https://***/pioneer-innovation/Zero-Shot-Object-Navigation. © 2022, CC BY.

关键词： Robots

Large Language Model-Driven Urban Traffic Signal control

学校读者我要写书评

暂无评论

Large Language Model-Driven Urban Traffic Signal Control

control Conference (ANZCC), Australian and New Zealand

作者： Yiqing Tang Xingyuan Dai Chen Zhao Qi Cheng Yisheng Lv School of Artificial Intelligence University of 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 School of Mechanical and Electrical Engineering China University of Mining and Technology at Beijing University of Mining and Technology (Beijing) Inner Mongolia Research Institute Ordos China

In recent years, large language models (LLM) have received a lot of attention for their ability to understand, generate and process natural language. By fine-tuning the models on specific domains or using prompts, LLMs can generate summaries tailored to different contexts and requirements. We hope to utilize these advantages of LLM in urban traffic signal control and provide a new paradigm for traffic control. In this paper, we propose a large language model-driven urban traffic control method, which is designed based on artificial systems, computational experiments, parallel execution (ACP) method. There are three traffic control modes, autonomous mode, feedback mode, and human takeover mode. In autonomous mode, we adopt recommendations based on inventory strategy, which can to some extent reduce the acquisition time of traffic control strategies. With the continuous construction and improvement of the strategic library, the efficiency of the system will be further improved. The feedback mode is mainly aimed at traffic scenarios that are not in the library, generating new control strategies and introducing human feedback. The human takeover model ensures that human engineers have direct control over the actual transportation system. LLM drives the interaction between human engineers and computers in the process of traffic control. Assist human engineers in achieving traffic control and improving work efficiency.

关键词：

Dual refinement networks for accurate and fast object detection in real-world scenes

学校读者我要写书评

暂无评论

arXiv 2018年

作者： Chen, Xingyu Yu, Junzhi Kong, Shihan Wu, Zhengxing Wen, Li State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 University of Chinese Academy of Sciences Beijing100049 China Beijing Innovation Center for Engineering Science and Advanced Technology Peking University Beijing100871 China School of Mechanical Engineering and Automation Beihang University Beijing100191 China

Object detection has been vigorously studied for years but fast accurate detection for real-world scenes remains a very challenging problem. Overcoming drawbacks of singlestage detectors, we take aim at precisely detecting objects for static and temporal scenes in real time. Firstly, as a dual refinement mechanism, a novel anchor-offset detection including an anchor refinement, a feature offset refinement, and a deformable detection head is designed for two-step regression and capture of accurate detection features. Based on the anchoroffset detection, a dual refinement network (DRN) is developed for high-performance static detection, where a multi-deformable head is further designed to leverage contextual information for describing objects. As for temporal detection in real-world scenes, temporal refinement networks (TRN) and temporal dual refinement networks (TDRN) are developed by propagating the refinement information across time, where we also propose a loose refinement strategy to match object motion with the previous refinement. Our proposed methods are evaluated on PASCAL VOC, COCO, and ImageNet VID datasets. Extensive comparison on static and temporal detection verify the superiority of DRN, TRN, and TDRN. Consequently, our developed approaches run in a fairly fast speed, and in the meantime achieve a significantly enhanced detection accuracy, i.e., a mAP of 82.8% on VOC 2007, 80.6% on VOC 2012, 69.4% on VID 2017, and an AP of 34.3% on COCO. Ultimately, producing encouraging results, our methods are applied to underwater object grasping with an autonomous system. Codes are publicly available at http://***/SeanChenxy/TDRN. Copyright © 2018, The Authors. All rights reserved.

关键词： Computer vision

TrajGen: Generating Realistic and Diverse Trajectories with Reactive and Feasible Agent Behaviors for Autonomous Driving

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhang, Qichao Gao, Yinfeng Zhang, Yikang Guo, Youtian Ding, Dawei Wang, Yunpeng Sun, Peng Zhao, Dongbin The State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China The College of Artificial Intelligence University of Chinese Academy of Sciences Beijing100049 China School of Automation and Electrical Engineering University of Science and Technology Beijing Beijing100083 China Baidu Inc. Beijing100085 China

Realistic and diverse simulation scenarios with reactive and feasible agent behaviors can be used for validation and verification of self-driving system performance without relying on expensive and time-consuming real-world testing. Existing simulators rely on heuristic-based behavior models for background vehicles, which cannot capture the complex interactive behaviors in real-world scenarios. To bridge the gap between simulation and the real world, we propose TrajGen, a two-stage trajectory generation framework, which can capture more realistic behaviors directly from human demonstration. In particular, TrajGen consists of the multi-modal trajectory prediction stage and the reinforcement learning based trajectory modification stage. In the first stage, we propose a novel auxiliary RouteLoss for the trajectory prediction model to generate multi-modal diverse trajectories in the drivable area. In the second stage, reinforcement learning is used to track the predicted trajectories while avoiding collisions, which can improve the feasibility of generated trajectories. In addition, we develop a data-driven simulator I-Sim that can be used to train reinforcement learning models in parallel based on naturalistic driving data. The vehicle model in I-Sim can guarantee that the generated trajectories by TrajGen satisfy vehicle kinematic constraints. Finally, we give comprehensive metrics to evaluate generated trajectories for simulation scenarios, which shows that TrajGen outperforms either trajectory prediction or inverse reinforcement learning in terms of fidelity, reactivity, feasibility, and diversity. Copyright © 2022, The Authors. All rights reserved.

关键词： Trajectories

Research and practice of lightweight digital twin speeding up the implementation of flexible manufacturing systems

学校读者我要写书评

暂无评论

Research and practice of lightweight digital twin speeding u...

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

作者： Xiaodong Zhang Bin Hu Gang Xiong Xinjie Liu Xisong Dong Dongjia Li School of Information Engineering Yantai institute of Technology Yantai China 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 Chinese Academy of Sciences Beijing China Yantai University Yantai China School of Artificial Intelligence Yantai Institute of Technology Yantai China

Parallel manufacturing in Industry 5.0 requires digital twin to digitize physical systems, building virtual models to open up channels connecting physical systems, information systems, and social systems, and transforming the physical models of the existing production environment to achieve two-way feedback of virtual and real is the current research direction. This paper proposes the modeling idea of lightweight digital twin, extracts core dimensions and performs digital virtual simulation, so as to quickly realize the complete process of two-way feedback, and realize a set of chess flexible parallel manufacturing production lines as a practice for the design of complete lightweight digital twin.

关键词： Industries Digital twin Conferences Buildings Production Manufacturing Flexible manufacturing systems

Enhanced biologically inspired model for image recognition based on a novel patch selection method with moment

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Lu, Yan-Feng Jia, Li-Hao Qaio, Hong Li, Yi Research Center for Brain-inspired Intelligence Institute of Automation Chinese Academy of Sciences Beijing China State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing China CAS Center for Excellence in Brain Science and Intelligence Technology Shanghai China School of Information Engineering Nanchang University Nanchang China

Biologically inspired model (BIM) for image recognition is a robust computational architecture, which has attracted widespread attention. BIM can be described as a four-layer structure based on the mechanisms of the visual cortex. Although the performance of BIM for image recognition is robust, it takes the randomly selected ways for the patch selection, which is sightless, and results in heavy computing burden. To address this issue, we propose a novel patch selection method with oriented Gaussian-Hermite moment (PSGHM), and we enhanced the BIM based on the proposed PSGHM, named as PBIM. In contrast to the conventional BIM which adopts the random method to select patches within the feature representation layers processed by multi-scale Gabor filter banks, the proposed PBIM takes the PSGHM way to extract a small number of representation features while offering promising distinctiveness. To show the effectiveness of the proposed PBIM, experimental studies on object categorization are conducted on the CalTech05, TU Darmstadt (TUD), and GRAZ01 databases. Experimental results demonstrate that the performance of PBIM is a significant improvement on that of the conventional BIM. Copyright © 2017, The Authors. All rights reserved.

关键词： Image enhancement

Heterogeneous Graph Reinforcement Learning for Dependency-aware Multi-task Allocation in Spatial Crowdsourcing

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhao, Yong Zhu, Zhengqiu Gao, Chen Wang, En Huang, Jincai Wang, Fei-Yue College of Systems Engineering National University of Defense Technology Hunan Province Changsha410073 China BNRist Tsinghua University Beijing100084 China College of Computer Science and Technology Jilin University Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun130012 China State Key Laboratory for Management and Control of Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100190 China

Spatial Crowdsourcing (SC) is gaining traction in both academia and industry, with tasks on SC platforms becoming increasingly complex and requiring collaboration among workers with diverse skills. Recent research works address complex tasks by dividing them into subtasks with dependencies and assigning them to suitable workers. However, the dependencies among subtasks and their heterogeneous skill requirements, as well as the need for efficient utilization of workers’ limited work time in the multi-task allocation mode, pose challenges in achieving an optimal task allocation scheme. Therefore, this paper formally investigates the problem of Dependency-aware Multi-task Allocation (DMA) and presents a well-designed framework to solve it, known as Heterogeneous Graph Reinforcement Learning-based Task Allocation (HGRL-TA). To address the challenges associated with representing and embedding diverse problem instances to ensure robust generalization, we propose a multi-relation graph model and a Compound-path-based Heterogeneous Graph Attention Network (CHANet) for effectively representing and capturing intricate relations among tasks and workers, as well as providing embedding of problem state. The task allocation decision is determined sequentially by a policy network, which undergoes simultaneous training with CHANet using the proximal policy optimization algorithm. Extensive experiment results demonstrate the effectiveness and generality of the proposed HGRL-TA in solving the DMA problem, leading to average profits that is 21.78% higher than those achieved using the metaheuristic methods. © 2024, CC BY-NC-ND.

关键词： Reinforcement learning

Acp-Net: Asymmetric Center Positioning Network for Real-Time Text Detection

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Zhu, Boyuan Liu, Fagui Chen, Xi Tang, Quan Chen, C.L. Philip South China University of Technology No.342 Outer Ring East Road Panyu Guangzhou510006 China Peng Cheng Laboratory No. 2 Xingke 1st Street Nanshan Shenzhen518055 China State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences No.95 Zhongguancun East Road Haidian Beijing100190 China

Scene text detection plays a crucial role in numerous application fields. However, despite the varying focus on real-time performance, almost all existing detection models employ the Feature Pyramid Network (FPN) structure for feature extraction, overlooking its inherent limitations. The integration of high-resolution multi-channel features within FPN demands significant computational resources. Although FPN which treats local and global features as equivalent, has been proven to be stable in various fields, it remains to be reconsidered whether this stability strategy is the most suitable for text features. To this end, we propose an Asymmetric Center Positioning Network (ACP-Net), which replaces the FPN structure and achieves precise and real-time text detection in complex scenarios. Specifically, we explore an asymmetric feature structure that includes two independent branches to acquire both global and local information, as well as an adaptive weighted fusion module to effectively capture the long-range dependencies from these two branches. Furthermore, we present a text center positioning module to enhance the understanding of text features by learning the feature centers. Comprehensive assessments conducted across various terminals have demonstrated the superiority of ACP-Net in terms of both accuracy and speed. © 2024, The Authors. All rights reserved.

关键词： Feature extraction

Modal regression based structured low-rank matrix recovery for multi-view learning

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Xu, Jiamiao Wang, Fangzhao Peng, Qinmu You, Xinge Wang, Shuo Jing, Xiao-Yuan Philip Chen, C.L. School of Electronic Information and Communications Huazhong University of Science and Technology Wuhan430074 China Shenzhen Huazhong University of Science and Technology Research Institute China State Key Laboratory of Software Engineering School of Computer Wuhan University China Department of Computer and Information Science Faculty of Science and Technology University of Macau 99999 China Dalian Maritime University Dalian116026 China State Key Laboratory of Management and Control for Complex Systems Institute of Automation Chinese Academy of Sciences Beijing100080 China

Low-rank Multi-view Subspace Learning (LMvSL) has shown great potential in cross-view classification in recent years. Despite their empirical success, existing LMvSL based methods are incapable of well handling view discrepancy and discriminancy simultaneously, which thus leads to the performance degradation when there is a large discrepancy among multi-view data. To circumvent this drawback, motivated by the block-diagonal representation learning, we propose Structured Low-rank Matrix Recovery (SLMR), a unique method of effectively removing view discrepancy and improving discriminancy through the recovery of structured low-rank matrix. Furthermore, recent low-rank modeling provides a satisfactory solution to address data contaminated by predefined assumptions of noise distribution, such as Gaussian or Laplacian distribution. However, these models are not practical since complicated noise in practice may violate those assumptions and the distribution is generally unknown in advance. To alleviate such limitation, modal regression is elegantly incorporated into the framework of SLMR (term it MR-SLMR). Different from previous LMvSL based methods, our MR-SLMR can handle any zero-mode noise variable that contains a wide range of noise, such as Gaussian noise, random noise and outliers. The alternating direction method of multipliers (ADMM) framework and half-quadratic theory are used to efficiently optimize MR-SLMR. Experimental results on four public databases demonstrate the superiority of MR-SLMR and its robustness to complicated noise. Copyright © 2020, The Authors. All rights reserved.

关键词： Gaussian noise (electronic)