Tesi etd-05172021-101901
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
BARBERI, FEDERICA
URN
etd-05172021-101901
Titolo
Naturalistic control of lower limb neuroprostheses optimizing intention decoding, feedback, and comfort.
Settore scientifico disciplinare
ING-IND/34
Corso di studi
Istituto di Biorobotica - BIOROBOTICS
Commissione
relatore MICERA, SILVESTRO
Parole chiave
- electromyography
- elettromiografia
- neuroprosthesis
- neurorobotics
- prosthesis
- prosthesis control
- protesi
- sensory feedback
Data inizio appello
30/06/2021;
Disponibilità
completa
Riassunto analitico
Subjects with lower-limb amputation present asymmetrical and unnatural gait. They usually implement compensation strategies that lead to reduced mobility and higher metabolic costs. By non-uniformly distributing their weight they often suffer from back pain and joint degeneration. Without sensory feedback, subjects with lower limb amputation experience a higher risk of falling, and they are less prone to rely on their prosthesis. The lack of sensations has a major role in the low embodiment of the prosthesis, which is perceived as a foreign body.
To face these issues, an optimized lower-limb prosthesis promoting a more natural and symmetric gait pattern is necessary. Users’ intended movements should be automatically detected without compromising the cognitive burden. By restoring sensory feedback, higher mobility, embodiment, and awareness of the surrounding environment would improve.
During my Ph.D. I contributed to three different aspects of lower-limb prostheses. First, a new controller for a semi-powered knee and ankle prosthesis was developed (Chapter 3). The controller aims at overcoming the limitations that commercial microprocessor lower-limb prostheses still have, especially during the swing phase of the gait. The controller was implemented to handle various locomotion modalities, such as level walking, stairs negotiation, ramp ascending, and ramp descending. Second, the control of a powered leg prosthesis based on the decoded intention of movement was studied (Chapter 4). We developed an intention decoding algorithm based on electromyographic (EMG) and inertial signals extracted from the amputated side of the subjects. The aim was to use a small number of EMG electrodes to ensure the highest possible comfort. The algorithm was tested offline and online on fourteen transfemoral and transtibial amputee subjects. With three EMG electrodes and three inertial sensors, the intention decoding algorithm achieved >95% classification performances. Third, we restored the sensory feedback in three transfemoral amputees using closed-loop and real-time intraneural stimulation (Chapter 5). Sensory feedback increased mobility, fall prevention, agility, and embodiment. We showed cognitive ease of the subjects when intraneural sensations were provided.
The developed controller (Chapter 3) for semi-powered prostheses showed potential to allow a closer-to-natural gait for amputees maintaining a high degree of controllability over the device. The intention-decoding algorithm presented in Chapter 4, allowed amputees to benefit from a prosthesis that automatically follows their intended movements without implementing strategies that could compromise their cognitive burden. Restored feedback sensations, illustrated in Chapter 5, ensure more confidence in amputees while walking, which in turn can lead to more even weight distribution and therefore symmetric gait.
Providing amputees with a comfortable set-up plays a key role with a view on commercializing the device. The results presented in this thesis help providing amputees with a functional but light experimental set-up for locomotion mode detection. It would be important to have this set-up embedded in the prosthesis and to develop a fully implantable system for the delivery of intraneural stimulation. Implementing the intention decoding algorithm on a semi-powered prosthesis, rather than a powered one, could also represent a potential way to overcome some limitations given by powered prosthetics, such as the heavy weight of the device, and its limited controllability. Overall, this work paves the way for future studies aiming at implementing new strategies for well-accepted lower-limb prostheses.
To face these issues, an optimized lower-limb prosthesis promoting a more natural and symmetric gait pattern is necessary. Users’ intended movements should be automatically detected without compromising the cognitive burden. By restoring sensory feedback, higher mobility, embodiment, and awareness of the surrounding environment would improve.
During my Ph.D. I contributed to three different aspects of lower-limb prostheses. First, a new controller for a semi-powered knee and ankle prosthesis was developed (Chapter 3). The controller aims at overcoming the limitations that commercial microprocessor lower-limb prostheses still have, especially during the swing phase of the gait. The controller was implemented to handle various locomotion modalities, such as level walking, stairs negotiation, ramp ascending, and ramp descending. Second, the control of a powered leg prosthesis based on the decoded intention of movement was studied (Chapter 4). We developed an intention decoding algorithm based on electromyographic (EMG) and inertial signals extracted from the amputated side of the subjects. The aim was to use a small number of EMG electrodes to ensure the highest possible comfort. The algorithm was tested offline and online on fourteen transfemoral and transtibial amputee subjects. With three EMG electrodes and three inertial sensors, the intention decoding algorithm achieved >95% classification performances. Third, we restored the sensory feedback in three transfemoral amputees using closed-loop and real-time intraneural stimulation (Chapter 5). Sensory feedback increased mobility, fall prevention, agility, and embodiment. We showed cognitive ease of the subjects when intraneural sensations were provided.
The developed controller (Chapter 3) for semi-powered prostheses showed potential to allow a closer-to-natural gait for amputees maintaining a high degree of controllability over the device. The intention-decoding algorithm presented in Chapter 4, allowed amputees to benefit from a prosthesis that automatically follows their intended movements without implementing strategies that could compromise their cognitive burden. Restored feedback sensations, illustrated in Chapter 5, ensure more confidence in amputees while walking, which in turn can lead to more even weight distribution and therefore symmetric gait.
Providing amputees with a comfortable set-up plays a key role with a view on commercializing the device. The results presented in this thesis help providing amputees with a functional but light experimental set-up for locomotion mode detection. It would be important to have this set-up embedded in the prosthesis and to develop a fully implantable system for the delivery of intraneural stimulation. Implementing the intention decoding algorithm on a semi-powered prosthesis, rather than a powered one, could also represent a potential way to overcome some limitations given by powered prosthetics, such as the heavy weight of the device, and its limited controllability. Overall, this work paves the way for future studies aiming at implementing new strategies for well-accepted lower-limb prostheses.
File
| Nome file | Dimensione |
|---|---|
| PhDThesi...final.pdf | 7.96 Mb |
Contatta l'autore |
|