A Hybrid Physical and Data-Driven Approach to Motion Prediction and Control in Human-Robot Collaboration /
Saved in:
| Author / Creator: | Wu, Min. |
|---|---|
| Imprint: | Berlin : Logos Verlag Berlin, [2022] |
| Description: | 1 online resource ( 220 pages) |
| Language: | English |
| Series: | Forschungsberichte Aus Dem Lehrstuhl Für Regelungssysteme ; band 22 Forschungsberichte aus dem Lehrstuhl für Regelungssysteme, Technische Universität Kaiserslautern ; bd. 22. |
| Subject: | |
| Format: | E-Resource Book |
| URL for this record: | http://pi.lib.uchicago.edu/1001/cat/bib/12967403 |
Table of Contents:
- Intro
- List of Tables
- List of Figures
- Notation
- Introduction
- Motivation
- Scenarios studied in this thesis
- State-of-the-art technologies
- Human modeling
- Robot control
- Design examples
- Objectives and Outline
- Publications
- Human Motion Prediction based on Simple Kinematic Models
- Introduction
- Data set preparation
- Mathematical formulations
- Constant acceleration model
- Minimum jerk model
- Evaluation and discussion
- Summary
- Data-driven Approaches for Human Motion Prediction
- Introduction
- Inverse optimal control
- Problem formulation
- Solution based on the principle of maximum entropy
- Implementation
- Evaluation and discussion
- Gaussian Process Regression
- Regular form
- Sparse Gaussian process
- Online Sparse Gaussian process
- Evaluation and discussion
- Gaussian process with explicit basis functions
- Summary
- Human Motion Prediction based on Dynamic Movement Primitives
- Introduction
- Hybrid physical and data-driven approach
- Modeling of interaction dynamics
- Regular DMP
- Modified DMP
- DMP with Gaussian process
- Rotation based spatial scaling
- Weighting factor
- Learning a DMP model
- Standard approach
- Two-step approach
- Evaluation and discussion
- Coupled DMP for modeling interaction dynamics
- Summary
- Discussion on Human Motion Prediction for Human-Robot Collaboration
- Comparison of different approaches in human motion prediction
- Boundary conditions
- Prediction accuracy at different time scales
- Implementation complexity and scope of application
- Potential research trends
- From single-predictor to multiple-predictor
- From simple motions to complex skills
- From individual to collaborative behaviors
- Physical Human-Robot Collaboration and Impedance Control
- Introduction
- Impedance control
- Classification of robot interaction control
- Joint impedance control
- Cartesian impedance control
- Analysis of physical human-robot collaboration
- System model
- Problem formulation
- Summary
- Adaptive Impedance Control based on Reinforcement Learning
- Introduction
- Basics of reinforcement learning
- Q-Learning-based adaptive impedance control
- Unconstrained Learning
- Learning with consideration of constraints
- Impedance and role assignment
- Summary
- An Adaptive Learning and Control Framework in Human-Robot Collaboration
- Introduction
- Experimental study: handovers
- Hardware
- Pre-alignment
- Design based on Gaussian Process
- Design based on Dynamic Movement Primitives
- Experimental study: object handling
- Validation of the Q-learning based adaptive impedance control
- Human reference estimation
- Evaluation of the whole framework
- Summary
- Conclusion and Outlook
- Conclusion
- Outlook
- Appendices
- Inverse Optimal Control with Local Optimality
- Bibliography
- Zusammenfassung
- Supervised theses
