Assistance Fonctionelle des mouvements de membres inférieurs par exosquelette

Functional assistance of lower limb movements using exoskeleton

• Auteur / Autrice : Rami Jradi
• Direction : Samer Mohammed
• Type : Projet de thèse
• Discipline(s) : Signal, Image, Automatique
• Date : Inscription en doctorat le ; Soutenance le 19/12/2024
• Etablissement(s) : Paris 12
• Ecole(s) doctorale(s) : École doctorale Mathématiques, Sciences et Technologies de l'Information et de la Communication
• Partenaire(s) de recherche : Laboratoire : LISSI - Laboratoire Images, Signaux et Systèmes Intelligents
• Jury : Président / Présidente : Christine Chevallereau
• Jury : Examinateurs / Examinatrices : Samer Mohammed, Antoine Ferreira, Jimmy Lauber, Yacine Amirat, Hala Rifai
• Jury : Rapporteurs / Rapporteuses : Antoine Ferreira, Jimmy Lauber

Mots clés

Mots clés libres

Assistance

Commande

Exosquelette

Exoskeleton

Assistance

Control

Résumé

Spinal cord injuries and stroke are major causes of motor disabilities, with foot drop being a common result of impaired neural communication between the central nervous system and the dorsiflexor muscles of the ankle joint. This impairment affects the ability to lift the foot during walking, leading to inefficient gait, increased fall risks, and reduced endurance. Actuated ankle-foot orthoses (AAFOs) aid in the rehabilitation of ankle disabilities by providing mechanical assistance, improving gait patterns, promoting natural joint movements, and enhancing stability. This thesis presents advanced control strategies for AAFOs that offer targeted assistance to individuals with foot drop by only providing the necessary assistance. These controllers are designed to improve the AAFO performance by adapting to the individual profile of each wearer and ensuring precise trajectory tracking, even in the presence of uncertainties or external disturbances. Trajectory tracking controllers based on active disturbance rejection control (ADRC) are proposed, with human muscular torque estimated through an observer. This enables the orthosis to complement the user’s effort in achieving the desired movement, guiding the ankle joint profile towards a healthy gait pattern. Unlike impedance and EMG-based controllers, the proposed approaches do not require residual effort or rely on EMG signals. The thesis introduces several contributions, beginning with an adaptive-based assist-as-needed control strategy, based on an adaptive active disturbance rejection control approach, the human muscular torque estimator is based first on fixed gains then on adaptive ones. Subsequently, a sigma-based adaptive ADRC method is integrated with machine learning algorithms to estimate ground reaction forces and detect gait sub-phases. Finally, a contraction-based ADRC is developed, focusing on simplifying implementation with reduced computational efforts and sensor requirements.