Daniela Cerrone
Prediction of dynamic data from kinematics captured by a depth camera-based telerehabilitation system.
Rel. Andrea Cereatti. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Biomedica, 2026
|
|
PDF (Tesi_di_laurea)
- Tesi
Accesso limitato a: Solo utenti staff fino al 28 Marzo 2028 (data di embargo). Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (32MB) |
|
|
Archive (ZIP) (Documenti_allegati)
- Altro
Accesso limitato a: Solo utenti staff fino al 28 Marzo 2028 (data di embargo). Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (15MB) |
Abstract
Telerehabilitation platforms have long lacked the ability to provide clinicians with real-time kinetic feedback, as obtaining Ground Reaction Forces and Joint Moments traditionally requires laboratory-grade force sensors unavailable in home settings. This work addresses that gap by integrating kinetic estimation directly into the Muvity telerehabilitation platform using only a depth camera. A multimodal dataset was collected from 25 subjects performing five functional movements and four clinical games. A MATLAB preprocessing pipeline handles signal synchronization, reference frame alignment, and musculoskeletal analysis through OpenSim Inverse Kinematics and Inverse Dynamics. Two deep learning pipelines were developed in Python and PyTorch. The first targets Ground Reaction Force estimation through three movement-specific families of physics-informed neural networks covering squat, balance, and stepping categories, each combining a dual-decoder MLP, adversarial subject-invariant training, and whole-body momentum constraints.
The second extends this approach to Joint Moment estimation, embedding bilateral symmetry, kinematic chain consistency, and temporal continuity constraints directly into the training loss
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Corso di laurea
Classe di laurea
Ente in cotutela
Aziende collaboratrici
URI
![]() |
Modifica (riservato agli operatori) |
