Jump to content
Learning-Based Inverse Dynamics for Human Motion Analysis / Titelei/Inhaltsverzeichnis
Learning-Based Inverse Dynamics for Human Motion Analysis / Titelei/Inhaltsverzeichnis
Contents
Chapter
Expand
|
Collapse
Page
Titelei/Inhaltsverzeichnis
Details
1–14
1 Introduction
1–14
Details
1.1 Applications and Challenges of Inverse Dynamics
Details
1.2 Learning Inverse Dynamics
Details
1.3 Contributions
Details
1.4 Structure of the Thesis
Details
1.5 Publications
Details
15–22
2 Related work
15–22
Details
2.1 Inverse Dynamics by Physical Simulation
Details
2.1.1 Inverse Approaches
Details
2.1.2 Forward Approaches
Details
2.1.3 Implicit Approaches
Details
2.2 Learning-Based Inverse Dynamics
Details
2.3 Decreasing Supervision
Details
23–61
3 Fundamentals
23–61
Details
3.1 Rigid Body Motion
Details
3.1.1 Representation of Position
Details
3.1.2 Representation of Orientation
Details
3.1.3 Homogeneous Transformations
Details
3.2 Kinematics of a Rigid Body System
Details
3.2.1 Kinematic Trees
Details
3.2.2 The Denavit-Hartenberg Convention
Details
3.2.3 Velocity and Acceleration Kinematics
Details
3.3 Dynamics of a Rigid Body System
Details
3.3.1 TMT-Method
Details
3.4 Machine Learning
Details
3.4.1 Terminology and General Concepts
Details
3.4.2 Support Vector Machines
Details
3.4.3 Ridge Regression
Details
3.4.4 Random Forests
Details
3.4.5 Neural Networks
Details
3.4.6 Generalization
Details
3.4.7 Transfer Learning
Details
62–75
4 Human motion dataset
62–75
Details
4.1 Motion Capture and Kinematic Optimization
Details
4.2 Force Plate Measurements
Details
4.3 Estimation of Inertial Properties
Details
4.4 Optimization of Joint Torques
Details
4.5 Data Specification
Details
4.6 Generation of Training Data Points
Details
76–94
5 Supervised learning of inverse dynamics
76–94
Details
5.1 Methodology
Details
5.1.1 End-to-End Regression
Details
5.1.2 Multi-Stage Regression
Details
5.2 Experimental Evaluation
Details
5.2.1 Predictive Dynamics Dataset
Details
5.2.2 Public Dataset
Details
5.2.3 Application to Reconstructed Motions
Details
5.3 Discussion
Details
95–122
6 Self-supervision by dynamics-based layers
95–122
Details
6.1 Datasets
Details
6.2 Dynamics Network
Details
6.2.1 Forward Layer
Details
6.2.2 Inverse Layer
Details
6.2.3 Contact Loss
Details
6.2.4 Training Modes
Details
6.3 Experimental Evaluation
Details
6.3.1 Comparison in the Supervised Setting
Details
6.3.2 Semi-Supervision with Small Labeled Datasets
Details
6.3.3 Domain Adaptation
Details
6.3.4 Ablation of Input Structure
Details
6.3.5 Effect of Noise
Details
6.4 Discussion
Details
123–126
7 Conclusions
123–126
Details
127–129
a Appendix
127–129
Details
a.1 Evaluation Based on Additional Metrics
Details
a.2 Data-Driven Inverse Dynamics Optimization
Details
130–152
Bibliography
130–152
Details
Durchsuchen Sie das Werk
Geben Sie ein Keyword in die Suchleiste ein
Partial access
Learning-Based Inverse Dynamics for Human Motion Analysis
Titelei/Inhaltsverzeichnis
Autoren
Petrissa Zell
DOI
doi.org/10.51202/9783186877109-I
ISBN print: 978-3-18-387710-2
ISBN online: 978-3-18-687710-9
Chapter Preview
Share
Download PDF
Download citation
RIS
BibTeX
Copy DOI link
doi.org/10.51202/9783186877109-I
Share by email
Video schließen
Share by email VDI Verlag eLibrary
Recipient*
Sender*
Message*
Your name
Send message
This site is protected by reCAPTCHA and the Google
Privacy Policy
and
Terms of Service
apply.