Sebastian Thrun - Böcker

Over the past three decades or so, research on machine learning and data mining has led to a wide variety of algorithms that learn general functions from experience. As machine learning is maturing, it has begun to make the successful transition from academic research to various practical applications. Generic techniques such as decision trees and artificial neural networks, for example, are now being used in various commercial and industrial applications. This text is a research direction within machine learning. Similar to traditional machine-learning algorithms, the methods described induce general functions from experience. However, the book investigates algorithms that can change the way they generalize, for example, practice the task of learning itself, and improve on it. To illustrate the utility of learning to learn, it is worthwhile comparing machine learning with human learning. Humans encounter a continual stream of learning tasks. They do not just learn concepts or motor skills, they also learn bias, meaning, they learn how to generalize.As a result, humans are often able to generalize correctly from extremely few examples - often just a single example suffices to teach us a new thing. A deeper understanding of computer programs that improve their ability to learn can have a large practical impact on the field of machine learning and beyond. In recent years, the field has made significant progress towards a theory of learning to learn along with practical new algorithms, some of which led to impressive results in real-world applications. The book's objective is to investigate the utility and feasibility of computer programs that can learn how to learn, both from a practical and a theoretical point of view.

Lifelong learning addresses situations in which a learner faces a series of different learning tasks providing the opportunity for synergy among them. Explanation-based neural network learning (EBNN) is a machine-learning algorithm that transfers knowledge across multiple learning tasks. When faced with a new learning task, EBNN exploits domain knowledge accumulated in previous learning tasks to guide generalization in the new one. As a result, EBNN generalizes more accurately from less data than comparable methods. This book describes the basic EBNN paradigm and investigates it in the context of supervised learning, reinforcement learning, robotics and chess.

This work contains seven papers on robot learning written by leading researchers in the field. As the selection of papers illustrates, the field of robot learning is both active and diverse. A variety of machine learning methods, ranging from inductive logic programming to reinforcement learning, is being applied to many subproblems in robot perception and control, often with objectives as diverse as parameter calibration and concept formulation. While no unified robot learning framework has yet emerged to cover the variety of problems and approaches described in these papers and other publications, a clear set of shared issues underlies many robot learning problems. These characteristics of robotics present challenges and constraints to the learning system. Since these characteristics are shared by other important real-world application domains, robotics is a highly attractive area for research on machine learning. On the other hand, machine learning is also highly attractive to robotics. There is a great variety of open problems in robotics that defy a static, hand-coded solution.

Lifelong learning addresses situations in which a learner faces a series of different learning tasks providing the opportunity for synergy among them. Explanation-based neural network learning (EBNN) is a machine learning algorithm that transfers knowledge across multiple learning tasks. When faced with a new learning task, EBNN exploits domain knowledge accumulated in previous learning tasks to guide generalization in the new one. As a result, EBNN generalizes more accurately from less data than comparable methods. Explanation-Based Neural Network Learning: A Lifelong Learning Approach describes the basic EBNN paradigm and investigates it in the context of supervised learning, reinforcement learning, robotics, and chess. `The paradigm of lifelong learning - using earlier learned knowledge to improve subsequent learning - is a promising direction for a new generation of machine learning algorithms. Given the need for more accurate learning methods, it is difficult to imagine a future for machine learning that does not include this paradigm.' From the Foreword by Tom M. Mitchell.

Over the past three decades or so, research on machine learning and data mining has led to a wide variety of algorithms that learn general functions from experience. As machine learning is maturing, it has begun to make the successful transition from academic research to various practical applications. Generic techniques such as decision trees and artificial neural networks, for example, are now being used in various commercial and industrial applications. Learning to Learn is an exciting new research direction within machine learning. Similar to traditional machine-learning algorithms, the methods described in Learning to Learn induce general functions from experience. However, the book investigates algorithms that can change the way they generalize, i.e., practice the task of learning itself, and improve on it. To illustrate the utility of learning to learn, it is worthwhile comparing machine learning with human learning. Humans encounter a continual stream of learning tasks. They do not just learn concepts or motor skills, they also learn bias, i.e., they learn how to generalize. As a result, humans are often able to generalize correctly from extremely few examples - often just a single example suffices to teach us a new thing. A deeper understanding of computer programs that improve their ability to learn can have a large practical impact on the field of machine learning and beyond. In recent years, the field has made significant progress towards a theory of learning to learn along with practical new algorithms, some of which led to impressive results in real-world applications. Learning to Learn provides a survey of some of the most exciting new research approaches, written by leading researchers in the field. Its objective is to investigate the utility and feasibility of computer programs that can learn how to learn, both from a practical and a theoretical point of view.

Recent Advances in Robot Learning contains seven papers on robot learning written by leading researchers in the field. As the selection of papers illustrates, the field of robot learning is both active and diverse. A variety of machine learning methods, ranging from inductive logic programming to reinforcement learning, is being applied to many subproblems in robot perception and control, often with objectives as diverse as parameter calibration and concept formulation. While no unified robot learning framework has yet emerged to cover the variety of problems and approaches described in these papers and other publications, a clear set of shared issues underlies many robot learning problems. Machine learning, when applied to robotics, is situated: it is embedded into a real-world system that tightly integrates perception, decision making and execution. Since robot learning involves decision making, there is an inherent active learning issue. Robotic domains are usually complex, yet the expense of using actual robotic hardware often prohibits the collection of large amounts of training data. Most robotic systems are real-time systems. Decisions must be made within critical or practical time constraints. These characteristics present challenges and constraints to the learning system. Since these characteristics are shared by other important real-world application domains, robotics is a highly attractive area for research on machine learning. On the other hand, machine learning is also highly attractive to robotics. There is a great variety of open problems in robotics that defy a static, hand-coded solution. Recent Advances in Robot Learning is an edited volume of peer-reviewed original research comprising seven invited contributions by leading researchers. This research work has also been published as a special issue of Machine Learning (Volume 23, Numbers 2 and 3).

This open access book is a practitioner's guide to smart, connected products and solutions. As a hands-on playbook, it combines the business and technical execution perspectives to help product companies, OEMs, manufacturers and equipment operators utilize the full potential of digital enablers, especially Artificial Intelligence (AI), Internet of Things (IoT) and Digital Twins. The Digital Playbook provides comprehensive and actionable guidance, helping to address the challenges of creating sustainable and scalable digital business models, managing cocreation and sourcing, setting up the digital organization, and handling the legal aspects. For the technical execution perspective, the playbook includes the AIoT Framework, which explains how to combine data science and AI engineering with Digital Twins, as well as software development for cloud and edge. The integration with physical product development and retrofit integration of existing equipment is included as well. A pragmatic, agile approach is introduced that takes common agile inhibitors into consideration. A holistic AIoT DevOps approach is described, which combines key elements of DevOps for cloud, edge and AI. Enterprise readiness is ensured by looking at trust and security as well as reliability and resilience for AIoT. A large number of real-world examples and case studies help ensure practical relevance.Readers should have a previous, general understanding of digital strategies and technologies. This book offers readers a clear understanding of the opportunities, as well as the challenges related to building and operating smart, connected products and solutions. They are given a set of tools and blueprints, which they can apply to their practical work in this space.

Since its inception in 1996, FSR, the biannual "International Conference on Field and Service Robotics" has published archival volumes of high reference value. This unique collection is the post-conference proceedings of the 4th FSR in Lake Yamanaka, Japan at July 2003. This book edited by Shin’ichi Yuta, Hajime Asama, Sebastian Thrun, Erwin Prassler and Takashi Tsubouchi is rich by topics and authoritative contributors and presents the current developments and new directions in field and service robotics. The contents of these contributions represent a cross-section of the current state of robotics research from one particular aspect: field and service applications, and how they reflect on the theoretical basis of subsequent developments. Pursuing technologies aimed at realizing skilful, smart, reliable, robust field and service robots is the big challenge running throughout this focused collection.

This monograph describes a new family of algorithms for the simultaneous localization and mapping problem in robotics (SLAM). SLAM addresses the problem of acquiring an environment map with a roving robot, while simultaneously localizing the robot relative to this map. This problem has received enormous attention in the robotics community in the past few years, reaching a peak of popularity on the occasion of the DARPA Grand Challenge in October 2005, which was won by the team headed by the authors. The FastSLAM family of algorithms applies particle filters to the SLAM Problem, which provides new insights into the data association problem that is paramount in SLAM. The FastSLAM-type algorithms have enabled robots to acquire maps of unprecedented size and accuracy, in a number of robot application domains and have been successfully applied in different dynamic environments, including the solution to the problem of people tracking.

Robotics is undergoing a major transformation in scope and dimension. From a largely dominant industrial focus, robotics is rapidly expanding into human environments and vigorously engaged in its new challenges. Interacting with, assisting, serving, and exploring with humans, the emerging robots will increasingly touch people and their lives. The Springer Tracts in Advanced Robotics (STAR) is devoted to bringing to the research community the latest advances in the robotics field on the basis of their significance and quality. Through a wide and timely dis­ semination of critical research developments in robotics, our objective with this series is to promote more exchanges and collaborations among the re­ searchers in the community and contribute to further advancements in this rapidly growing field. As one of robotics pioneering symposia, the International Symposium on Robotics Research (ISRR) has established over the past two decades some of the fields most fundamental and lasting contributions. Since the launching of STAR, ISRR and several other thematic symposia in robotics find an important platform for closer links and extended reach within the robotics community. This twelfth edition of Robotics Research, edited by Sebastian Thrun, Rodney Brooks, and Hugh Durrant-Whyte, offers in its 14-part volume a collection of a broad range of topics in robotics. The content of these contributions provides a wide coverage of the current state of robotics research: the advances and challenges in its theoretical foundation and technology basis, and the developments in its traditional and novel areas of apphcations.

This monograph describes a new family of algorithms for the simultaneous localization and mapping problem in robotics (SLAM). SLAM addresses the problem of acquiring an environment map with a roving robot, while simultaneously localizing the robot relative to this map. This problem has received enormous attention in the robotics community in the past few years, reaching a peak of popularity on the occasion of the DARPA Grand Challenge in October 2005, which was won by the team headed by the authors. The FastSLAM family of algorithms applies particle filters to the SLAM Problem, which provides new insights into the data association problem that is paramount in SLAM. The FastSLAM-type algorithms have enabled robots to acquire maps of unprecedented size and accuracy, in a number of robot application domains and have been successfully applied in different dynamic environments, including the solution to the problem of people tracking.

Robotics is undergoing a major transformation in scope and dimension. From a largely dominant industrial focus, robotics is rapidly expanding into human environments and vigorously engaged in its new challenges. Interacting with, assisting, serving, and exploring with humans, the emerging robots will increasingly touch people and their lives. The Springer Tracts in Advanced Robotics (STAR) is devoted to bringing to the research community the latest advances in the robotics field on the basis of their significance and quality. Through a wide and timely dis­ semination of critical research developments in robotics, our objective with this series is to promote more exchanges and collaborations among the re­ searchers in the community and contribute to further advancements in this rapidly growing field. As one of robotics pioneering symposia, the International Symposium on Robotics Research (ISRR) has established over the past two decades some of the fields most fundamental and lasting contributions. Since the launching of STAR, ISRR and several other thematic symposia in robotics find an important platform for closer links and extended reach within the robotics community. This twelfth edition of Robotics Research, edited by Sebastian Thrun, Rodney Brooks, and Hugh Durrant-Whyte, offers in its 14-part volume a collection of a broad range of topics in robotics. The content of these contributions provides a wide coverage of the current state of robotics research: the advances and challenges in its theoretical foundation and technology basis, and the developments in its traditional and novel areas of apphcations.