Intelligent Gait Analysis using Marker Based Motion Capturing System
- Marker-based systems can digitally record human movements in detail. Using the digital biomechanical human model Dynamicus, which was developed by the Institut für Mechatronik, it is possible to model joint angles and their velocities such accurately that it can be used to improve motion analysis in competitive sports or for ergonomic evaluation of motion sequences. In this paper, we use interpretable machine learning techniques to analyze the gait. Here, the focus is on the classification between foot touchdown and drop-off during normal walking. The motion data for training the model is labeled using force plates. We analyze how we could apply our machine learning models directly on new motion data recorded in a different scenario compared to the initial training, more precise on a treadmill. We use the properties of the interpretable model to detect drift and to transfer our model if necessary.
Author: | Danny Möbius, Marika Kaden, Daniel Staps, Thomas Villmann |
---|---|
URN: | urn:nbn:de:bsz:mit1-opus4-122947 |
DOI: | https://doi.org/10.48446/opus-12294 |
ISSN: | 1437-7624 |
Parent Title (German): | 26. Interdisziplinäre Wissenschaftliche Konferenz Mittweida |
Publisher: | Hochschule Mittweida |
Place of publication: | Mittweida |
Document Type: | Conference Proceeding |
Language: | English |
Year of Completion: | 2021 |
Publishing Institution: | Hochschule Mittweida |
Contributing Corporation: | Institut für Mechatronik Chemnitz e.V. |
Release Date: | 2021/05/18 |
GND Keyword: | Motion Capturing; Ganganalyse; Künstliche Intelligenz |
Issue: | 002 |
Page Number: | 4 |
First Page: | 133 |
Last Page: | 136 |
Open Access: | Frei zugänglich |
Licence (German): | Urheberrechtlich geschützt |