Knowledge-Based Vision-Guided Robots

1 Introduction.- 1.1 Background.- 1.1.1 A vision-guided approach.- 1.1.2 Computer vision and vision-guided mobile robots.- 1.1.3 Applying high-level computer vision to guide mobile robots.- 1.2 Aims of the Research Presented in this Book: A Problem in Robot Vision.- 1.3 The Approach of this Book.- 1.4 About the Chapters.- 2 Related Systems and Ideas.- 2.1 Basic computer vision approaches.- 2.1.1 Frame-based computer vision.- 2.1.2 Active vision.- 2.2 Vision-Guided Mobile Robot Systems.- 2.2.1 Mobile robot subsystems and concepts.- 2.2.2 Mobile robot object recognition.- 2.2.3 Maps and path planning.- 2.2.4 Temporal sequencing for complex tasks.- 2.2.5 Vision-guided mobile robot systems.- 2.2.6 Reactive navigation.- 2.2.7 Model-based vision systems for mobile robots.- 2.2.8 Knowledge-based mobile robotic systems.- 2.2.9 Vision-guided mobile robots using stereo.- 2.2.10 Active perception systems for mobile robots.- 2.2.11 Application of vision-guided mobile robots.- 2.3 Computer Vision for Mobile Robots.- 2.3.1 Traditional model-based vision 3D object recognition.- 2.3.2 Shape-from-shading.- 2.3.3 Pose determination.- 2.4 Conclusion.- 3 Embodied Vision For Mobile Robots.- 3.1 Introduction.- 3.1.1 Embodiment.- 3.1.2 Phenomena and noumena.- 3.2 The Classical Computer Vision Paradigm.- 3.2.1 Non-classical computer vision.- 3.3 Problems with Classical Computer Vision.- 3.4 Applying Embodied Concepts in Human Vision.- 3.4.1 Models play an analogous role in computer vision.- 3.5 Embodiment of Vision-guided Robots.- 3.5.1 Embodiment, task and environment.- 3.5.2 The role of the task.- 3.5.3 The role of the environment.- 3.6 Embodiment for Vision-guided Robots.- 3.6.1 Physical embodiment.- 3.6.2 Embodiment in a task.- 3.6.3 Embodiment in an environment.- 3.7 Conclusion.- 4 Object Recognition Mobile Robot Guidance.- 4.1 Introduction.- 4.2 System Perspective.- 4.3 Object Recognition.- 4.3.1 Canonical-views.- 4.3.2 Match verification.- 4.3.3 Edge matching.- 4.3.4 Edge-based features for ground-based robots.- 4.3.5 View prediction.- 4.4 Determining Object Pose and Distance.- 4.4.1 Active determination of the sign of ?.- 4.4.2 Error analysis.- 4.5 Conclusion.- 5 Edge Segmentation and Matching.- 5.1 Edge Extraction.- 5.1.1 Edge extraction.- 5.1.2 On the choice of window size and quantisation of ? and ?.- 5.2 Edge Matching.- 5.2.1 Evaluating matches.- 5.2.2 Spatial elimination.- 5.2.3 Edge coverage.- 5.2.4 Position estimation consistency.- 5.2.5 Geometric verification.- 5.2.6 Quadratic edge extraction.- 5.2.7 Further active processing.- 6 Knowledge Based Shape from Shading.- 6.1 Introduction.- 6.1.1 Motivation and system perspective.- 6.1.2 Assumptions.- 6.1.3 Knowledge-based representation of objects.- 6.2 Using Object Model Knowledge for Shape-From-Shading.- 6.3 A New Boundary Condition for Shape-From-Shading.- 6.4 Knowledge-based Implementation.- 6.4.1 Knowledge / frame topology.- 6.4.2 Fact knowledge.- 6.4.3 Procedural knowledge.- 6.4.4 Shape processing rulebase.- 6.5 Experimental Method and Results.- 6.5.1 Synthetic images.- 6.5.2 Real images.- 6.5.3 Domain knowledge.- 6.6 Conclusion.- 7 Supporting Navigation Components.- 7.1 Model-based Path Planning.- 7.1.1 Path planning and obstacle avoidance.- 7.2 Odometry and Obstacle Avoidance Subsystem.- 7.2.1 Obstacle avoidance strategies.- 7.2.2 Coordinate transforms.- 8 Fuzzy Control for Active Perceptual Docking.- 8.1 Introduction.- 8.1.1 Fuzzy control.- 8.1.2 Fuzzy control for mobile robot control.- 8.1.3 TSK fuzzy model.- 8.1.4 Visual motion-based approaches to mobile robots and the docking problem.- 8.2 Direction Control for Robot Docking.- 8.2.1 The log-polar camera.- 8.2.2 Docking for a ground-based robot.- 8.2.3 Noise in the input parameter.- 8.3 A Fuzzy Control Scheme.- 8.4 Results.- 8.5 Conclusion.- 9 System Results and Case Studies.- 9.1 Evaluation of Components.- 9.1.1 Experimental setup.- 9.1.2 View matching.- 9.1.3 Pose determination — power supply.- 9.1.4 Pose determination — model vehicle.- 9.2 Case Studies.- 9.2.1 Moving around the corner of an object.- 9.2.2 Distinguishing a particular object among similar objects.- 9.2.3 Docking.- 9.2.4 Object circumnavigation.- 9.2.5 Obstacle avoidance.- 9.3 Conclusion.- 10 Conclusion.- 10.1 Limitations of the Research Presented and Future Work.- 10.2 Extended quotation from Descartes.