دانلود کتاب مدل‌سازی صحنه سه‌بعدی و تعامل رباتیک

General Secretary Xi Jinping pointed out in the report of the 20th National Congress of the Communist Party of China: Education, science and technology, and human resources are the foundational and strategic pillars for building a modern socialist country in all respects. We must regard science and technology as our primary productive force, talent as our primary resource, and innovation as our primary driver of growth. We will fully implement the strategy for invigorating China through science and education, the workforce development strategy, and the innovation-driven development strategy. We will open up new areas and new arenas in development and steadily foster new growth drivers and new strengths. The report also emphasizes: We will promote the integrated and clustered development of strategic emerging industries and cultivate new growth engines such as next-generation information technology, artificial intelligence, biotechnology, new energy, new materials, highend equipment, and green industry. At present, the overall advancement of the development of new generation artificial intelligence-related disciplines, theoretical modeling, technological innovation, hardware and software upgrades, etc., is triggering chain breakthroughs, accelerating the transition of various fields of economy and society from digitization and networking to intelligence. General Secretary Xi Jinping pointed out at the Academicians Conference of the Chinese Academy of Sciences and the Academicians Conference of the Chinese Academy of Engineering that robots are the “the pearl on the crown of the manufacturing industry”. Embodied intelligent robots, as carriers of core technology in the field of artificial intelligence, possess the ability to interact with the environment like humans, autonomously plan, decide, act, and execute, playing an important role in serving the country’s major needs, leading the national economic development and ensuring national defense security. Perceiving the 3D environment and understanding the information contained therein is an important manifestation of human intelligence and a prerequisite for free interaction between humans and the environment. Similarly, intelligent robots also need to first fully perceive the environment they are in, which essentially involves 3D digital modeling and understanding of the surrounding scene. The National “14th Five-Year Plan for the Development of the Robot Industry” explicitly states the need to strengthen the cultivation of robotics talent. At present, although some books involve some of the technologies, there is no reference book that systematically and comprehensively introduces the above fields and connects them into a cohesive whole. This book focuses on the intelligent perception and interaction module in the intelligent robot system, establishes a multidisciplinary cross-integration knowledge system, from the perspective of scene modeling and understanding, discusses the technical frontier and research frontier as comprehensively as possible, and carries out practical discussions on robot navigation, obstacle avoidance, and grasping and other practical application tasks. This book combines theory and practice, covering both basic algorithms and the latest techniques, with detailed code or pseudo-code resources. This book consists of 6 chapters. Chapters 1–3 comprehensively introduce the development and application of scene modeling and understanding technology, including 3D geometry and semantic reconstruction, scene active exploration, and scene structure understanding. Chapter 1 introduces 3D geometry and semantic reconstruction algorithms based on traditional multi-view geometry and deep learning, discussing how to recover 3D structure from images. Chapter 2 introduces scene active exploration algorithms in 3D reconstruction and visual semantic navigation, discussing different ways to explore unknown scenes. Chapter 3 introduces two types of scene structure understanding algorithms based on object detection and classification and scene graphs, discussing different ways of scene structure understanding from the two levels of unstructured and structured scene knowledge. Chapters 4 and 5 comprehensively introduce the development and application of robot perception technology, including visual relocalization and robot navigation, obstacle avoidance, grasping, etc. Chapter 4 introduces classic and deep learningbased algorithms related to visual relocalization, navigation, and obstacle avoidance, discussing how robots acquire their own pose status and complete navigation and obstacle avoidance tasks. Chapter 5 introduces analytical methods, empirical methods, and deep learning-based algorithms for robot grasping and interaction, discussing different grasping posture detection and robot grasping methods. Chapter 6 focuses on 3D scene understanding for robot interaction tasks in practice, which is convenient for readers to comprehensively understand and master theory and practice. As for the content of this book, except for Chapter 6, each chapter corresponds to a fixed theme, in which the theory and practice parts are interspersed. In each chapter, the theme is first outlined, the main research work at home and abroad is discussed around this theme, and different technologies are explained according to different methods. The explanation of technology is mainly practical, combined with theoretical knowledge in practice, and “chapter knowledge points” and other characteristic content are designed to facilitate readers to deepen their understanding; each technology includes task description, related work, technical methods, results presentation and summary, etc. The relevant content materials (algorithm code, related papers, etc.) of this book are attached in the form of the SN-branded ShortURLs in each chapter. Tong Li, Xuefeng Yin, Yuanbo Wang, Yuanyuan Liu, Jiajin Qiu, Hu Lin, Zhaoxuan Zhang, Boyan Wei, Zijian Xu, Xiangji Cheng, Yunduo Zhou, Qirui Bao, Xiaochen Shang, Shanglai Qu, Jianchuan Ding, and others from National and Local Joint Engineering Laboratory of Computer Aided Design at Dalian University of Technology participated in the partial organization of this manuscript. Some of the research results in this book originate from National 863 Project “Robot Technology as a Networked and Cloud Computing Terminal,” National Natural Science Foundation of China’s key project “Visual and Semantic-based Indoor Scene Understanding and Real-Time Modeling,” Ministry of Science and Technology’s “Brain Science and Brain-like Research” Major Program “Design and Application of Multi-modal and Multi-scale Synergistic Neuroplasticity Brain-like Reinforcement Learning Network Model,” National Natural Science Foundation of China’s Major Research Program cultivation project “Research on Natural Interaction Methods for Co-Integrated Robots for Dynamic Unstructured Environments,” National Natural Science Foundation of China’s General Program “Research on Fast Visual Re-localization Methods for Building Interior Scenes under High-speed and Complex Lighting Environments,” and Dalian Jieqing Program “Intelligent Sensing and Decision Making System for Bionic Robots.” This book can serve as a teaching reference book in the fields of information and intelligence related majors, computer graphics, computer vision, and intelligent robots in higher education institutions, and can also be used as a reference for professionals in related fields. In the field of artificial intelligence, the development of scene modeling and robot perception technology is advancing rapidly. Although this book strives to be comprehensive and closely follow its development trend, due to the limited level and time of the authors, there are inevitably omissions in the book. Readers are kindly encouraged to provide feedback and corrections.