Lower-limb exoskeletons have emerged as promising solutions for gait assistance and rehabilitation, yet their performance strongly depends on accurate and robust motion sensing. In most wearable systems, kinematic estimation is based on inertial measurement units (IMUs), which are widely adopted but remain affected by drift, sensitivity to placement, and the need for recurrent calibration. In this thesis, a soft hip exosuit exploiting magnetic sensing for motion tracking is presented and experimentally validated as an alternative to conventional inertial sensing. The proposed system integrates a bilateral soft exosuit with a magnetic tracking architecture based on an array of magnetic field sensors and a permanent magnet attached to the user's leg. The estimated thigh motion is processed through an Extended Kalman Filter to reconstruct gait phase and gait frequency in real time, enabling phase-synchronized generation of assistive motor commands. The work addresses the complete development pipeline, including sensing setup design, control architecture implementation, exosuit assembly, and experimental validation. First, a kinematic evaluation was conducted on eight participants walking on a treadmill at different speeds in order to compare magnetic-tracking-based estimation with IMU-derived measurements. The results showed a strong agreement between the two approaches, with only minimal timing differences between the estimated gait phases and high correlation in gait-frequency estimation (group-level Pearson correlation coefficients of 0.9793 ± 0.0115 for the left leg and 0.9829 ± 0.0092 for the right leg). The proposed framework also generated stable and repeatable reference trajectories, demonstrating that magnetic tracking can reliably support real-time exosuit control. Second, a metabolic evaluation was performed to assess the functional effectiveness of the device during walking. The assisted condition yielded the lowest metabolic cost, with statistically significant reductions relative to both the unassisted exosuit condition (p = 0.011) and normal walking without the device (p = 0.038). Overall, this work demonstrates that magnetic tracking is a feasible and effective sensing modality for soft exoskeleton control. By enabling accurate, drift-free gait-state estimation and synchronized assistance, the proposed approach represents a promising alternative to IMU-based sensing for future wearable assistive systems.
Gli esoscheletri per gli arti inferiori sono emersi come soluzioni promettenti per l’assistenza al cammino e la riabilitazione; tuttavia, le loro prestazioni dipendono fortemente da una stima del movimento accurata e robusta. Nella maggior parte dei sistemi indossabili, la stima cinematica si basa su unità di misura inerziali (IMU), ampiamente adottate ma ancora soggette a deriva, sensibilità al posizionamento e necessità di calibrazioni ricorrenti. In questa tesi viene presentato e validato sperimentalmente un esosuit morbido per l’anca che sfrutta il sensing magnetico per il motion tracking, come alternativa al sensing inerziale convenzionale. Il sistema proposto integra un esosuit morbido bilaterale con un’architettura di tracciamento magnetico basata su un array di sensori di campo magnetico e su un magnete permanente fissato alla gamba dell’utente. Il movimento stimato della coscia viene elaborato mediante un filtro di Kalman esteso per ricostruire in tempo reale la fase del passo e la frequenza del cammino, consentendo la generazione sincronizzata in fase dei comandi motori di assistenza. Il lavoro affronta l’intera pipeline di sviluppo, includendo la progettazione del sistema di sensing, l’implementazione dell’architettura di controllo, l’assemblaggio dell’esosuit e la validazione sperimentale. In primo luogo, è stata condotta una valutazione cinematica su otto partecipanti che hanno camminato su tapis roulant a differenti velocità, al fine di confrontare la stima basata sul tracciamento magnetico con le misure derivate dalle IMU. I risultati hanno mostrato una forte concordanza tra i due approcci, con differenze temporali minime tra le fasi del passo stimate e un’elevata correlazione nella stima della frequenza del cammino (coefficienti di correlazione di Pearson a livello di gruppo pari a 0.9793 ± 0.0115 per la gamba sinistra e 0.9829 ± 0.0092 per la gamba destra). Il framework proposto ha inoltre generato traiettorie di riferimento stabili e ripetibili, dimostrando che il tracciamento magnetico può supportare in modo affidabile il controllo in tempo reale dell’esosuit. In secondo luogo, è stata effettuata una valutazione metabolica per analizzare l’efficacia funzionale del dispositivo durante il cammino. La condizione assistita ha mostrato il costo metabolico più basso, con riduzioni statisticamente significative sia rispetto alla condizione con esosuit non assistito (p = 0.011) sia rispetto al cammino naturale senza dispositivo (p = 0.038). Nel complesso, questo lavoro dimostra che il tracciamento magnetico rappresenta una modalità di sensing fattibile ed efficace per il controllo di esoscheletri morbidi. Consentendo una stima accurata dello stato del cammino, priva di deriva, e un’assistenza sincronizzata, l’approccio proposto costituisce una promettente alternativa al sensing basato su IMU per i futuri sistemi indossabili di assistenza.
Esoscheletro Morbido per il Supporto dell'Anca con Sensori Magnetici per la Rilevazione Cinematica
DE RIVO, VALENTINO
2025/2026
Abstract
Lower-limb exoskeletons have emerged as promising solutions for gait assistance and rehabilitation, yet their performance strongly depends on accurate and robust motion sensing. In most wearable systems, kinematic estimation is based on inertial measurement units (IMUs), which are widely adopted but remain affected by drift, sensitivity to placement, and the need for recurrent calibration. In this thesis, a soft hip exosuit exploiting magnetic sensing for motion tracking is presented and experimentally validated as an alternative to conventional inertial sensing. The proposed system integrates a bilateral soft exosuit with a magnetic tracking architecture based on an array of magnetic field sensors and a permanent magnet attached to the user's leg. The estimated thigh motion is processed through an Extended Kalman Filter to reconstruct gait phase and gait frequency in real time, enabling phase-synchronized generation of assistive motor commands. The work addresses the complete development pipeline, including sensing setup design, control architecture implementation, exosuit assembly, and experimental validation. First, a kinematic evaluation was conducted on eight participants walking on a treadmill at different speeds in order to compare magnetic-tracking-based estimation with IMU-derived measurements. The results showed a strong agreement between the two approaches, with only minimal timing differences between the estimated gait phases and high correlation in gait-frequency estimation (group-level Pearson correlation coefficients of 0.9793 ± 0.0115 for the left leg and 0.9829 ± 0.0092 for the right leg). The proposed framework also generated stable and repeatable reference trajectories, demonstrating that magnetic tracking can reliably support real-time exosuit control. Second, a metabolic evaluation was performed to assess the functional effectiveness of the device during walking. The assisted condition yielded the lowest metabolic cost, with statistically significant reductions relative to both the unassisted exosuit condition (p = 0.011) and normal walking without the device (p = 0.038). Overall, this work demonstrates that magnetic tracking is a feasible and effective sensing modality for soft exoskeleton control. By enabling accurate, drift-free gait-state estimation and synchronized assistance, the proposed approach represents a promising alternative to IMU-based sensing for future wearable assistive systems.| File | Dimensione | Formato | |
|---|---|---|---|
|
DeRivo_Valentino.pdf
embargo fino al 14/04/2027
Dimensione
7.3 MB
Formato
Adobe PDF
|
7.3 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/106814