Venue: Centre Broca and on Zoom
Defense in french
Effie Segas
Team: Hybrid (INCIA)
Directeur de thèse : M Aymar de Rugy, Directeur de recherche (INCIA)
Title
Contrôle biomimétique de prothèses à partir des mouvements résiduels et d’informations contextuelles.
Biomimetic control of prostheses based on residual movements and contextual information.
Abstract
Although limb deficiency could be considered as a rare condition, it strongly impacts the quality of life of the persons suffering from it. Despite advances in myoelectric prostheses, whose control scheme is based on the electrical activity of residual muscles, the lack of relevant control signals for a trans-humeral level of disability does not allow for simultaneous and intuitive movement of multiple arm joints necessary to bring the hand to various positions and orientations needed to grasp objects. Grounded on advances in movement-based prosthesis control, we recently proposed a promising alternative using predictions from an artificial neural network, receiving the position and orientation of the movement goal as well as the current angular position of the residual joints, to control the degrees of freedom lost following a trans-humeral amputation. In the first set of experiments presented in this thesis, this control is adapted to be used by an amputee. To do so, the entire upper limb movements of 10 able-bodied participants, performed while reaching targets in various positions and orientations, are used to train a new neural network, designed to be usable by an amputee. In a virtual environment, the use of this network adapted to their morphology allowed 19 participants, including 7 amputees, to reach various targets without prior training with very high success rates (> 99 %) and reaching times comparable to those of natural movement. Using this same approach, 15 participants, including 2 acquired and 1 congenital amputees, controlled a robotic arm to grasp real objects with high success rates and reaching times consistent with natural movement, in a simplified context where neither trunk nor shoulder movements could be used to compensate for control imperfections. This approach allows a good convergence of the hand on the targeted object, which offers direct possibility of application in virtual environment for the treatment of phantom limb pain. However, it implies an abrupt modification of the distal configuration at each change of target that should be better handled for a real application. In the second part of this thesis, a new approach is tested to eliminate this abrupt change by proposing a smooth transition, determined from the speed of the stump movements and the gap between the current and the “goal” distal configurations. Two methods are presented to define this “goal” configuration, either from the position and orientation of the movement goal (i.e. target object) alone or by also taking into account the current orientation of the stump. In virtual reality, these controls ensure a smooth transition with, however, an extended movement time, and allowed 12 able-bodied participants to reach targets at various positions and orientations. The good results obtained, particularly when the current orientation of the stump is taken into account, allow to consider a transition on a real device whose proof of principle is brought in the last study of this thesis. Despite performances below those observed in virtual reality, which can be explained by the presence of various constraints (e.g. mechatronics, management of discontinuity, non-egocentric point of view, addition of compensatory movements), the 12 valid participants of this experiment were able to grasp most of the proposed targets. These results point to the need for further work to improve the discontinuity management and the real test experimental setup.
Key words : movement-based control, trans-humeral prosthesis, amputee, artificial neural network, motor coordination, joint synergies.
Publications
MICK, S., SEGAS, E., DURE, L., HALGAND, C., BENOIS-PINEAU, J., LOEB, G. E., CATTAERT, D., & de RUGY, A. (2021).
Shoulder kinematics plus contextual target information enable control of multiple distal joints of a simulated prosthetic arm and hand [Publisher : BioMed Central]. Journal of NeuroEngineering and Rehabilitation, 18(1). https://doi.org/10.1186/s12984-020-00793-0
SEGAS, E., MICK, S., LECONTE, V., KLOTZ, R., CATTAERT, D., & de RUGY, A. (In press).
Intuitive movement-based prosthesis control enables arm amputees to reach naturally in virtual reality (eLife). Rehabilitation Medicine et Physical Therapy.
preprint (2022): https://doi.org/10.1101/2022.10.15.22281053
Jury
Présidente : Mme Hélène Sauzéon, Professeure des universités (Centre Inria de l’université de Bordeaux)
Rapporteur : M Nathanaël Jarassé, Chargé de recherche (CNRS / ISIR)
Rapporteur : M Vincent Padois, Directeur de Recherche (Centre Inria de l’université de Bordeaux)
Examinatrice : Mme Jozina De graaf, Maîtresse de conférences (Université Aix-Marseille)
Examinateur : M Michel Guerraz, Professeur des universités (Université Savoie Mont-Blanc)
Invité : M Rémi Klotz, Docteur (Tour de Gassies)
Directeur de thèse : M Aymar de Rugy, Directeur de recherche (INCIA)