Tesi etd-03092021-113021
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
GUAITOLINI, MICHELANGELO
URN
etd-03092021-113021
Titolo
Body inertial sensor networks and multiple sensor fusion techniques for human motion studies
Settore scientifico disciplinare
ING-INF/06
Corso di studi
Istituto di Biorobotica - BIOROBOTICS
Commissione
relatore Prof. SABATINI, ANGELO MARIA
Parole chiave
- body sensor networks
- inertial sensors
- kinematics
- Motion analisys
- motor control
- sensor fusion
- stability
Data inizio appello
15/06/2021;
Disponibilità
completa
Riassunto analitico
Typical sensing techniques for human motion analysis rely on camera-based systems or wearable sensors. While camera-based systems still represent the standard in terms of accuracy, their high cost and their being limited to indoor analysis limit their use. On the other hand, wearable sensors may represent a promising technology. MEMS technology allowed to produce low cost, small sensors. Tri-axial accelerometers, gyroscope and magnetometers are currently among the most used MEMS and they are often combined in magnetic inertial measurement units (MIMUs).
However, MIMUs are error-prone sensors, needing for computational methods to compensate inaccuracies. To solve this problem, multiple data sources (namely, multiple MIMUs) may be fused in a body sensors network (BSN) to obtain redundant or complementary information, necessary to get both a more complete and a more accurate perspective of the motion task under exam. For this research project MIMU-based BSNs will be considered and any further reference to sensors constituting a BSN will imply that MIMUs are employed.
BSN represent one of the most promising solution for the future of motion analysis, and this is true in a wide variety of fields. However they require sophisticated signal processing techniques and their design remain a challenge that with this work was our intent to face. So, the aim of this research project is to investigate the potential of MIMU-based BSNs in motion analysis, spanning from healthcare and assistance to sport applications.
The potential of BSN in motion analysis, with a focus on stability, kinematics reconstruction and in sports will be investigated proposing setups and algorithms that may better describe motions under observation. Furthermore, a BSNs may give enough spatial coverage on user’s body to allow for kinematics reconstruction; body segment constraints and proper sensor fusion algorithms may represent a way to increase kinematics reconstruction accuracy, offering a versatile alternative to camera-based systems.
However, MIMUs are error-prone sensors, needing for computational methods to compensate inaccuracies. To solve this problem, multiple data sources (namely, multiple MIMUs) may be fused in a body sensors network (BSN) to obtain redundant or complementary information, necessary to get both a more complete and a more accurate perspective of the motion task under exam. For this research project MIMU-based BSNs will be considered and any further reference to sensors constituting a BSN will imply that MIMUs are employed.
BSN represent one of the most promising solution for the future of motion analysis, and this is true in a wide variety of fields. However they require sophisticated signal processing techniques and their design remain a challenge that with this work was our intent to face. So, the aim of this research project is to investigate the potential of MIMU-based BSNs in motion analysis, spanning from healthcare and assistance to sport applications.
The potential of BSN in motion analysis, with a focus on stability, kinematics reconstruction and in sports will be investigated proposing setups and algorithms that may better describe motions under observation. Furthermore, a BSNs may give enough spatial coverage on user’s body to allow for kinematics reconstruction; body segment constraints and proper sensor fusion algorithms may represent a way to increase kinematics reconstruction accuracy, offering a versatile alternative to camera-based systems.
File
| Nome file | Dimensione |
|---|---|
| Guaitoli...Final.pdf | 3.04 Mb |
Contatta l'autore |
|