Tesi etd-03282023-091655
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
IANNICIELLO, VALERIO
URN
etd-03282023-091655
Titolo
Towards a Myokinetic Interface for Prosthetic Control
An Embedded Real-Time Multi-Magnet Localizer
Settore scientifico disciplinare
ING-IND/34
Corso di studi
Istituto di Biorobotica - PHD IN BIOROBOTICA
Commissione
relatore Prof. CIPRIANI, CHRISTIAN
Membro Prof.ssa CASADIO, MAURA
Membro Dott. D'ALONZO, MARCO
Membro Prof.ssa CASADIO, MAURA
Membro Dott. D'ALONZO, MARCO
Parole chiave
- myokinetic interface
- upper limb prosthetic
- embedded electronics
- prosthesis control
- magnetic localization
Data inizio appello
03/07/2023;
Disponibilità
parziale
Riassunto analitico
Developing a dexterous hand prosthesis that can be controlled and perceived by the amputee in a natural manner is a major challenge in bioengineering. Recent years have seen the rapid development and evo-lution of surgical techniques and technologies aimed at this purpose, the majority of which probes the electrical activity of muscles for controlling the prosthesis and delivers electrical pulses to nerves for partially restor-ing sensory feedback. However, a human-machine interface able to al-low a biomimetic, bi-directional connection between the brain and the prosthesis is still to be achieved.
This challenge was undertaken and the idea of the myokinetic interface was introduced. It leverages the muscle movements of the residual limb for controlling the robotic hand and mechanical stimuli for the sensory feedback, mirroring the natural way a hand is controlled through an in-novative application of magnetic field principles. Small permanent mag-nets are implanted in independent muscles and thus move together with them, allowing the monitoring of their contractions and enabling the corresponding movements in the artificial hand to be controlled in a biomimetic, direct, independent, and parallel manner. Additionally, by positioning external coils on the socket, magnets can be selectively vi-brated at physiological frequencies to convey kinesthetic percepts.
To achieve this goal, an embedded multi-magnet localizer that can be seamlessly integrated into a prosthetic socket was designed and real-ized. Such system can track the position and the orientation of an arbi-trary number of magnets properly positioned, at frequencies up to 100Hz. It features a modular architecture that allows it to adapt to the varying needs and specifications of different amputations. The self-contained multi-magnet localizer, the main topic of the present disserta-tion, was designed over the past years in parallel to all the components needed for the final implantation of the myokinetic interface. Indeed, upcoming clinical trials will allow the assessment, for the first time, of the system feasibility with amputees. In this context, the outcomes of the localizer (i.e., the magnet position and orientation in space) will be provided as input to different kinds of control algorithms (e.g., direct control or pattern recognition), implemented on the same embedded system. Thus, the one presented here is intended to be exploited as a fully integrated prosthetic control system, which at present has only been preliminary tested in the laboratory environment.
In view of the clinical translation of the interface, some of the po-tential issues that may arise when transferring this system to a real-world scenario have also been identified. Magnetic and electromagnetic inter-ference produced by common objects can compromise the accuracy of the localizations, and therefore their effect on localization was investigat-ed to analyze the need of shielding the system in different conditions. Moreover, mechanical disturbances caused by external forces acting on the socket, are likely to induce unwanted relative displacements between the sensors and magnets or between the magnets themselves. To ad-dress this issue, some corrective algorithms were designed and tested.
Finally, the system was preliminarily tested with two patients who underwent targeted muscle reinnervation (TMR) surgery, in a non-invasive manner. TMR typically results in a significant physical displace-ment of the reinnervated muscles during contraction, which is a draw-back for the use of surface electrodes. The system was adapted for a non-invasive approach which exploits this effect by anchoring the mag-nets to the skin surface, and it was shown that six different voluntary movements could effectively be discriminated using a simple logistic regressor.
The system is going to be soon evaluated, for the first time, on a volunteer with a transradial amputation. Six magnets will be implanted in two flexor and one extensor muscles, and a fully integrated myokinetic prosthetic system will be used to assess the effectiveness of the ap-proach with respect to the golden standard myoelectric control. Fur-thermore, this opportunity will be exploited to measure the ability of vibrating magnets to provide proprioceptive feedback.
The work detailed in this thesis significantly contributed to the de-velopment of the myokinetic interface, opening new possibilities for humans to interface with robotic technologies in a more intuitive and natural way.
This challenge was undertaken and the idea of the myokinetic interface was introduced. It leverages the muscle movements of the residual limb for controlling the robotic hand and mechanical stimuli for the sensory feedback, mirroring the natural way a hand is controlled through an in-novative application of magnetic field principles. Small permanent mag-nets are implanted in independent muscles and thus move together with them, allowing the monitoring of their contractions and enabling the corresponding movements in the artificial hand to be controlled in a biomimetic, direct, independent, and parallel manner. Additionally, by positioning external coils on the socket, magnets can be selectively vi-brated at physiological frequencies to convey kinesthetic percepts.
To achieve this goal, an embedded multi-magnet localizer that can be seamlessly integrated into a prosthetic socket was designed and real-ized. Such system can track the position and the orientation of an arbi-trary number of magnets properly positioned, at frequencies up to 100Hz. It features a modular architecture that allows it to adapt to the varying needs and specifications of different amputations. The self-contained multi-magnet localizer, the main topic of the present disserta-tion, was designed over the past years in parallel to all the components needed for the final implantation of the myokinetic interface. Indeed, upcoming clinical trials will allow the assessment, for the first time, of the system feasibility with amputees. In this context, the outcomes of the localizer (i.e., the magnet position and orientation in space) will be provided as input to different kinds of control algorithms (e.g., direct control or pattern recognition), implemented on the same embedded system. Thus, the one presented here is intended to be exploited as a fully integrated prosthetic control system, which at present has only been preliminary tested in the laboratory environment.
In view of the clinical translation of the interface, some of the po-tential issues that may arise when transferring this system to a real-world scenario have also been identified. Magnetic and electromagnetic inter-ference produced by common objects can compromise the accuracy of the localizations, and therefore their effect on localization was investigat-ed to analyze the need of shielding the system in different conditions. Moreover, mechanical disturbances caused by external forces acting on the socket, are likely to induce unwanted relative displacements between the sensors and magnets or between the magnets themselves. To ad-dress this issue, some corrective algorithms were designed and tested.
Finally, the system was preliminarily tested with two patients who underwent targeted muscle reinnervation (TMR) surgery, in a non-invasive manner. TMR typically results in a significant physical displace-ment of the reinnervated muscles during contraction, which is a draw-back for the use of surface electrodes. The system was adapted for a non-invasive approach which exploits this effect by anchoring the mag-nets to the skin surface, and it was shown that six different voluntary movements could effectively be discriminated using a simple logistic regressor.
The system is going to be soon evaluated, for the first time, on a volunteer with a transradial amputation. Six magnets will be implanted in two flexor and one extensor muscles, and a fully integrated myokinetic prosthetic system will be used to assess the effectiveness of the ap-proach with respect to the golden standard myoelectric control. Fur-thermore, this opportunity will be exploited to measure the ability of vibrating magnets to provide proprioceptive feedback.
The work detailed in this thesis significantly contributed to the de-velopment of the myokinetic interface, opening new possibilities for humans to interface with robotic technologies in a more intuitive and natural way.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|