This book introduces a novel model-based dexterous manipulation framework, which, thanks to its precision and versatility, significantly advances the capabilities of robotic hands compared to the previous state of the art. This is achieved by combining a novel grasp state estimation algorithm, the first to integrate information from tactile sensing, proprioception and vision, with an impedance-based in-hand object controller, which enables leading manipulation capabilities, including finger gaiting. The developed concept is implemented on one of the most advanced robotic manipulators, the DLR humanoid robot David, and evaluated in a range of challenging real-world manipulation scenarios and tasks. This book greatly benefits researchers in the field of robotics that study robotic hands and dexterous manipulation topics, as well as developers and engineers working on industrial automation applications involving grippers and robotic manipulators.
Author(s): Martin Pfanne
Series: Springer Tracts in Advanced Robotics, 149
Publisher: Springer
Year: 2022
Language: English
Pages: 212
City: Cham
Series Editor’s Foreword
Preface
Contents
List of Symbols and Abbreviations
List of Symbols
Abbreviations
List of Figures
List of Tables
1 Introduction
1.1 Dexterous Manipulation
1.2 Contribution
1.3 Organization of this Work
References
2 Related Work
2.1 Dexterous Robotic Hands
2.1.1 A Brief History of Robotic Hands
2.1.2 DLR David
2.2 Dexterous Manipulation
2.2.1 Overview
2.2.2 Grasp State Estimation
2.2.3 Impedance-Based Object Control
2.2.4 Learning-Based Methods
References
3 Grasp Modeling
3.1 Definitions
3.2 Kinematics
3.2.1 Forward Kinematics
3.2.2 Grasp Matrix
3.2.3 Hand Jacobian
3.2.4 Contact Model
3.3 Dynamics
3.3.1 Rigid Body Dynamics
3.3.2 Grasp Dynamics
3.3.3 Contact Dynamics
3.4 Grasp Subspaces
3.5 Types of Grasps
References
4 Grasp State Estimation
4.1 Introduction
4.1.1 Concept
4.1.2 Problem Statement
4.2 Probabilistic Grasp State Estimation
4.2.1 Fundamentals
4.2.2 Particle Filter
4.2.3 Extended Kalman Filter
4.2.4 Filter Selection
4.3 Contact Detection and Localization
4.3.1 Collision Detection
4.3.2 Joint Torque Measurements
4.3.3 Contact Point Localization
4.4 State Estimation from Finger Position Measurements
4.4.1 Grasp State Definition
4.4.2 Motion Model
4.4.3 Measurement Model
4.4.4 Extensions
4.5 Data Fusion with Fiducial Markers
4.5.1 AprilTag
4.5.2 Measurement Model
4.5.3 Camera Localization
4.5.4 Target Tracking
4.6 Data Fusion with Contour Features
4.6.1 Feature Extraction
4.6.2 Measurement Model
4.7 Data Fusion with Visual Object Tracking
4.7.1 Multi-Modality Visual Object Tracking
4.7.2 Measurement Model
4.8 Data Fusion Under Measurement Delays
4.9 Experimental Validation
4.9.1 Grasp Acquisition
4.9.2 Pick-and-Place
4.9.3 In-Hand Manipulation
4.10 Summary
References
5 Impedance-Based Object Control
5.1 Introduction
5.1.1 Concept
5.1.2 Problem Statement
5.2 Controller Design
5.2.1 Object Impedance
5.2.2 Force Distribution
5.2.3 Architecture Overview
5.3 Object Impedance Control
5.3.1 Object Positioning
5.3.2 Maintaining the Grasp Configuration
5.4 Internal Forces
5.4.1 Force Distribution
5.4.2 Quadratic Optimization
5.4.3 Extensions
5.5 Torque Mapping
5.5.1 Force Mapping
5.5.2 Nullspace Control
5.6 Grasp Reconfiguration
5.6.1 Adding and Removing Contacts
5.6.2 Grasp Acquisition
5.7 Enabling In-Hand Manipulation
5.7.1 Finger Gaiting Interface
5.7.2 Contact Point Relocation
5.8 Experimental Validation
5.8.1 Tracking Performance
5.8.2 Stabilizing the Grasp Acquisition
5.8.3 Finger Gaiting
5.9 Summary
References
6 Conclusion
6.1 Summary and Discussion
6.2 Outlook
References
