Inhalt

•  
   Abstract
•  
   Contents
•  
   1 Motivation
•  
   1.1 Background
•  
   1.2 Aim
•  
   1.3 Robot Skills Needed for Gestural Understanding
•  
   1.4 Functionalities for Realizing Gesture Understanding
•  
   1.4.1 Detection of the Hand
•  
   1.4.2 Tracking of the Hand
•  
   1.4.3 Recognition of the Gesture
•  
   1.4.4 Incorporating Context for Understanding the Gesture
•  
   1.4.5 Beyond Communication: Recognizing Manipulative Actions
•  
   1.5 Terminology
•  
   1.6 Organization of the Book
•  
  2 Gestures in Human-Robot Interaction
•  
   2.1 Categorizations of Gestures
•  
   2.1.1 General Categorizations
•  
   2.1.2 Gestures in Human-Computer Interaction
•  
   2.1.3 A Gesture Categorization for Human-Robot Interaction
•  
   2.1.4 Selected Categories for Creating Intelligent Robots
•  
   2.1.5 The Influence of Context on Gesture Understanding
•  
   2.2 Sensing Devices for Observing Gesturing Humans
•  
   2.2.1 Intrusive Sensing Methods
•  
   2.2.2 Active Sensors
•  
   2.2.3 Vision-based Sensing Methods
•  
   2.2.4 Choosing the Right Sensor
•  
   2.3 Design Decisions for Gesture Understanding Systems
•  
   2.3.1 Graylevel vs. Color Images
•  
   2.3.2 Data-driven vs. Model-driven Processing(Bottom-up vs. Top-down)
•  
   2.3.3 Modular vs. Holistic Approaches
•  
   2.3.4 Gesture Recognition vs. Hand Detection/Recognition
•  
   2.4 Summary
•  
  3 Detection of the Hand
•  
   3.1 Hand Detection vs. Posture Recognition
•  
  3.2 Modeling the Hand's Visual Features
•  
   3.2.1 Hand Shape
•  
   3.2.2 Skin Color
•  
   3.2.3 Hand Appearance
•  
  3.3 Model-based Hand Detection
•  
   3.3.1 Explicit Modeling of the Hand Constituents
•  
   3.3.2 Holistic Models of the Hand
•  
   3.4 Summary and Conclusion
•  
  4 Tracking of the Hand
•  
   4.1 Detection vs. Adaptation
•  
  4.2 Tracking based on Hand Detection
•  
   4.2.1 Kalman Filtering
•  
   4.2.2 Particle Filtering
•  
   4.3 Adaptive Visual Features for Hand Tracking
•  
   4.3.1 Adaptive Hand Color
•  
   4.3.2 Adaptive Hand Shape
•  
   4.3.3 Adaptive Hand Appearance
•  
  4.4 Example: Detecting and Tracking Hands Based on Skin Color
•  
   4.4.1 System Overview
•  
   4.4.2 Modeling Skin Color Distribution and Skin Locus
•  
   4.4.3 Applying Skin Color Segmentation
•  
   4.4.4 Updating the Skin Color Model
•  
   4.4.5 Evaluation Results
•  
  4.5 Model-based Approaches to Hand Tracking
•  
   4.5.1 Model-based Tracking of Hand Configurations
•  
   4.5.2 Model-based Tracking of Arm/Body Configurations
•  
   4.6 Example: Tracking 3D Human Body Configurations
•  
   4.6.1 System Overview
•  
   4.6.2 Modeling the Appearance of Humans
•  
   4.6.3 Tracking Multiple Body Configuration Hypotheses
•  
   4.6.4 Evaluation Results
•  
   4.7 Summary and Conclusion
•  
   5 Recognition of the Gesture
•  
  5.1 Holistic Methods Applying Implicit Models of Hand Gestures
•  
   5.1.1 Single-Frame Gesture Recognition
•  
   5.1.2 Holistic Temporal Models
•  
   5.2 Modular Methods for Matching Trajectories: General Design
•  
   5.3 Deterministic Matching Methods
•  
   5.3.1 Direct Comparison
•  
   5.3.2 Dynamic Time Warping
•  
   5.3.3 Modeling Sequences of Atomic Gestures
•  
  5.4 Probabilistic Approaches for Trajectory Matching
•  
   5.4.1 Hidden-Markov-Models for Trajectory Recognition
•  
   5.4.2 Hierarchical HMMs and Dynamic Bayesian Networks
•  
   5.4.3 Particle Filtering for Trajectory Recognition
•  
   5.4.4 Trajectory Matching Approaches employing Neural Networks
•  
  5.5 Example: Trajectory-Based Recognition of Pointing Gestures
•  
   5.5.1 System Overview
•  
   5.5.2 Recognizing Pointing Gestures
•  
   5.5.3 Evaluation Results
•  
   5.6 Summary and Conclusion
•  
   6 Incorporating Context for Understanding the Gesture
•  
   6.1 Incorporating User-Provided Context for Pointing Gestures
•  
   6.1.1 Posture Information Restricting the Object Search Space
•  
   6.1.2 Verbal Information Complementing Pointing Gestures
•  
   6.1.3 Verbal Information Specifying Object Properties
•  
  6.2 Example: Including Verbal Cues for Resolving Object References
•  
   6.2.1 System Overview
•  
   6.2.2 Finding Previously Known Objects
•  
   6.2.3 Learning Views of Unknown Objects
•  
   6.2.4 Evaluation Results
•  
   6.3 Incorporating Situational Context for Manipulative Gestures
•  
   6.3.1 Body-Centered Action Recognition
•  
   6.3.2 Object-Centered Action Recognition
•  
   6.3.3 Parallel Approaches Combining Objects and Gestures
•  
   6.3.4 Holistic Approaches to Action Recognition
•  
  6.4 Example: A Fusion Approach for Recognizing Manipulative Actions
•  
   6.4.1 System Overview
•  
   6.4.2 Inferring Process
•  
   6.4.3 Evaluation Results
•  
   6.5 Summary and Conclusion
•  
  7 Robots Exhibiting Gesture Understanding Capabilities
•  
   7.1 Robots Understanding Communicative Gestures
•  
   7.1.1 Symbolic and Conventional Gestures
•  
   7.1.2 Referential and Pointing Gestures
•  
   7.2 Robots Understanding Manipulative Actions
•  
   7.2.1 Imitating the Observed Motion
•  
   7.2.2 Understanding for Reacting to the Environment Manipulation
•  
   7.3 Summary and Conclusion
•  
  8 Towards Learning of Gestures:Studies on Human Gesture Understanding
•  
   8.1 Limits of Classical Approaches to Learning
•  
   8.2 Modifications of Child-directed Motions: Motionese
•  
   8.3 Technical Analysis of Motionese
•  
   8.4 Studying Gestural Interaction between Humans and a Robot
•  
   8.5 Summary and Conclusion
•  
   9 Conclusion
•  
   References