Learning-Based Robot Vision

✍ Scribed by Josef Pauli

• Publisher: Springer
• Year: 2001
• Tongue: English
• Leaves: 292
• Category: Library

✦ Synopsis

Industrial robots carry out simple tasks in customized environments for which it is typical that nearly all e?ector movements can be planned during an - line phase. A continual control based on sensory feedback is at most necessary at e?ector positions near target locations utilizing torque or haptic sensors. It is desirable to develop new-generation robots showing higher degrees of autonomy for solving high-level deliberate tasks in natural and dynamic en- ronments. Obviously, camera-equipped robot systems, which take and process images and make use of the visual data, can solve more sophisticated robotic tasks. The development of a (semi-) autonomous camera-equipped robot must be grounded on an infrastructure, based on which the system can acquire and/or adapt task-relevant competences autonomously. This infrastructure consists of technical equipment to support the presentation of real world training samples, various learning mechanisms for automatically acquiring function approximations, and testing methods for evaluating the quality of the learned functions. Accordingly, to develop autonomous camera-equipped robot systems one must ?rst demonstrate relevant objects, critical situations, and purposive situation-action pairs in an experimental phase prior to the application phase. Secondly, the learning mechanisms are responsible for - quiring image operators and mechanisms of visual feedback control based on supervised experiences in the task-relevant, real environment. This paradigm of learning-based development leads to the concepts of compatibilities and manifolds. Compatibilities are general constraints on the process of image formation which hold more or less under task-relevant or accidental variations of the imaging conditions.

✦ Table of Contents

Learning-Based Robot Vision
Preface
Contents
1. Introduction
1.1 Need for New-Generation Robot Systems
Present State of Robotics
Problems and Requirements in Robotics
New Application Areas for Camera-Equipped Robot Systems
Contribution and Novelty of This Book
1.2 Paradigms of Computer Vision (CV) and Robot Vision (RV)
1.2.1 Characterization of Computer Vision
1.2.2 Characterization of Robot Vision
1.3 Robot Systems versus Autonomous Robot Systems
1.3.1 Characterization of a Robot System
1.3.2 Characterization of an Autonomous Robot System
1.3.3 Autonomous Camera-Equipped Robot System
1.4 Important Role of Demonstration and Learning
1.4.1 Learning Feature Compatibilities under Real Imaging
1.4.2 Learning Feature Manifolds of Real World Situations
1.4.3 Learning Environment-Effector-Image Relationships
1.4.4 Compatibilities, Manifolds, and Relationships
1.5 Chapter Overview of the Work
2. Compatibilities for Object Boundary Detection
2.1 Introduction to the Chapter
2.1.1 General Context of the Chapter
2.1.2 Object Localization and Boundary Extraction
2.1.3 Detailed Review of Relevant Literature
2.1.4 Outline of the Sections in the Chapter
2.2 Geometric/Photometric Compatibility Principles
2.2.1 Hough Transformation for Line Extraction
2.2.2 Orientation Compatibility between Lines and Edges
2.2.3 Junction Compatibility between Pencils and Corners
2.3 Compatibility-Based Structural Level Grouping
2.3.1 Hough Peaks for Approximate Parallel Lines
2.3.2 Phase Compatibility between Parallels and Ramps
2.3.3 Extraction of Regular Quadrangles
2.3.4 Extraction of Regular Polygons
2.4 Compatibility-Based Assembly Level Grouping
2.4.1 Focusing Image Processing on Polygonal Windows
2.4.2 Vanishing-Point Compatibility of Parallel Lines
2.4.3 Pencil Compatibility of Meeting Boundary Lines
2.4.4 Boundary Extraction for Approximate Polyhedra
2.4.5 Geometric Reasoning for Boundary Extraction
2.5 Visual Demonstrations for Learning Degrees of Compatibility
2.5.1 Learning Degree of Line/Edge Orientation Compatibility
2.5.2 Learning Degree of Parallel/Ramp Phase Compatibility
2.5.3 Learning Degree of Parallelism Compatibility
2.6 Summary and Discussion of the Chapter
3. Manifolds for Object and Situation Recognition
3.1 Introduction to the Chapter
3.1.1 General Context of the Chapter
3.1.2 Approach for Object and Situation Recognition
3.1.3 Detailed Review of Relevant Literature
3.1.4 Outline of the Sections in the Chapter
3.2 Learning Pattern Manifolds with GBFs and PCA
3.2.1 Compatibilitya nd Discriminabilityf or Recognition
3.2.2 Regularization Principles and GBF Networks
3.2.3 Canonical Frames with Principal Component Analysis
3.3 GBF Networks for Approximation of Recognition Functions
3.3.1 Approach of GBF Network Learning for Recognition
3.3.2 Object Recognition under ArbitraryVi ew Angle
3.3.3 Object Recognition for ArbitraryV iew Distance
3.3.4 Scoring of Grasping Situations
3.4 Sophisticated Manifold Approximation for Robust Recognition
3.4.1 Making Manifold Approximation Tractable
3.4.2 Log-Polar Transformation for Manifold Simplification
3.4.3 Space-Time Correlations for Manifold Refinement
3.4.4 Learning Strategywi th PCA/GBF Mixtures
3.5 Summary and Discussion of the Chapter
4. Learning-Based Achievement of RV Competences
4.1 Introduction to the Chapter
4.1.1 General Context of the Chapter
4.1.2 Learning Behavior-Based Systems
4.1.3 Detailed Reviewof Relevant Literature
4.1.4 Outline of the Sections in the Chapter
4.2 Integrating Deliberate Strategies and Visual Feedback
4.2.1 Dynamical Systems and Control Mechanisms
4.2.2 Generic Modules for System Development
4.3 Treatment of an Exemplary High-Level Task
4.3.1 Description of an Exemplary High-Level Task
4.3.2 Localization of a Target Object in the Image
4.3.3 Determining and Reconstructing Obstacle Objects
4.3.4 Approaching and Grasping Obstacle Objects
4.3.5 Clearing Away Obstacle Objects on a Parking Area
4.3.6 Inspection and/or Manipulation of a Target Object
4.3.7 Monitoring the Task-Solving Process
4.3.8 Overall Task-Specific Configuration of Modules
4.4 Basic Mechanisms for Camera–Robot Coordination
4.4.1 Camera–Manipulator Relation for One-Step Control
4.4.2 Camera–Manipulator Relation for Multi-step Control
4.4.3 Hand Servoing for Determining the Optical Axis
4.4.4 Determining the Field of Sharp View
4.5 Summary and Discussion of the Chapter
5. Summary and Discussion
5.1 Developing Camera-Equipped Robot Systems
5.2 Rationale for the Contents of This Work
5.3 Proposals for Future Research Topics
Appendix 1: Ellipsoidal Interpolation
Appendix 2: Further Behavioral Modules
Symbols
Index
References