- Titel
- NaturalWalk: An Anatomy-based Synthesizer for Human Walking Motions
- Zitierfähige Url:
- https://nbn-resolving.org/urn:nbn:de:bsz:ch1-qucosa2-854197
- Übersetzter Titel (DE)
- NaturalWalk: Ein Anatomie basierter Generator für menschliche Gehbewegungen
- Schriftenreihe
- Chemnitzer Informatik-Berichte
- Bandnummer
- CSR-23-05
- ISSN
- 0947-5125
- Abstract (EN)
- We present a novel data-driven approach for synthesizing human gait motions with individual style characteristics and natural appearance. Our approach is based on the concept of a motion signature that captures the essential characteristic of an individual walking motion. For each joint angle our motion model consists of a shape template and feature functions that describe the variation of that shape with the stride length. For the synthesis of a walking motion, the feature functions are evaluated for a desired stride length. Then the templates are adapted to match the computed features and used as progressions for the joint angles of the skeleton. We demonstrate our data driven approach using motion data captured from 12 individuals. We report on an experiment showing that the synthesized motions have a natural appearance and maintain the individual style.
- Andere Ausgabe
- Chemnitzer Informatik-Berichte
Link: http://www.tu-chemnitz.de/informatik/service/ib - Freie Schlagwörter (EN)
- Motion capturing, Walking motion synthesis, Individual motion characteristics, Template-based motion synthesis, Motion perception
- Klassifikation (DDC)
- 000
- Normschlagwörter (GND)
- Motion capturing, Walking
- Herausgeber (Institution)
- Technische Universität Chemnitz
- Technische Universität Chemnitz
- Technische Universität Chemnitz
- Technische Universität Chemnitz
- Version / Begutachtungsstatus
- eingereichte Version / Preprint
- URN Qucosa
- urn:nbn:de:bsz:ch1-qucosa2-854197
- Veröffentlichungsdatum Qucosa
- 10.05.2023
- Dokumenttyp
- Forschungsbericht
- Sprache des Dokumentes
- Englisch
- Lizenz / Rechtehinweis
CC BY 4.0- Inhaltsverzeichnis
1. Introduction 2. Related Work 3. Preliminaries 3.1 Mathematics of motion 3.2 Walking motions 4. Data acquisition and analysis 5. Shape templates and feature functions 5.1 Definition of template functions 5.2 Continuous representation of template functions 5.3 Building the feature functions 6. Motion Generation 6.1 Adaption of template functions 6.2 Computing the poses 7 Experimental Results 7.1 Numerical Evaluation 7.2 User Study Acknowledgment References