Gaëtan Courtois' thesis defense

Hemiplegia, mainly caused by stroke, results in a motor and/or sensory deficit on one side of the body, making walking difficult or even impossible. To help hemiplegic patients regain their ability to walk, therapists recommend specific exercises in the early stages of rehabilitation, involving the joints of the paralyzed lower limb. In view of the increase in strokes, the limited number of therapists and the need to intensify rehabilitation, it is necessary to identify complementary rehabilitation solutions.

At present, lower-limb exoskeletons in rehabilitation focus mainly on repeating gait cycles, but do not offer diversified exercises adapted to hemiplegic patients in the image of those practiced in rehabilitation centers. The general objective, within which this research work falls, is to develop a gait rehabilitation exoskeleton for the severe hemiplegic patient, which aims to fill this identified gap.

This research work aims to identify the mechatronic means and develop the control laws to assist mobilization of the hemiplegic patient's lower limb in the sagittal plane. The choices and developments made are based on a synthesis of elements from biomechanics, relating to the locomotor apparatus and the gait cycle, and rehabilitation, both conventional and robotic. It turns out that the exoskeleton must satisfy three main constraints: it must be ergonomic and guarantee the safety of both patient and therapist; it must provide adequate motor torque to assist the patient's lower limb in its movements; and it must offer modulation of assistance to encourage the patient's active participation and thus promote recovery.

To meet these constraints, the mechatronic means (actuation system and instrumentation) of a first hip-actuated exoskeleton prototype, as well as the associated control system, are proposed. The mechatronics comprise: a brushless DC motor, a deformation-wave gearbox, an inductive encoder at the motor output, and a torque sensor at the gearbox output. The control system developed is based on a hybrid control law enabling the transition between a control law based on a linear impedance model with constant parameters, dedicated to transparency, and a control law based on an impedance model with variable stiffness and damping parameters, which enables the assistance to be modulated in real time according to the patient's participation in task completion.

The impedance control laws used calculate the required assistance torque as a function of a reference trajectory. Online trajectory generation algorithms, are proposed to reproduce a trajectory close to those followed by a therapist during rehabilitation tasks. These algorithms are used to generate reference trajectories for three rehabilitation tasks defined for use with the exoskeleton: a hip flexion-extension task; a cadenced hip flexion-extension task; and a gait cycle repetition task. The trajectories generated by the algorithms obtained in simulation are comparable to measurements obtained experimentally on a half-dozen asymptomatic subjects manipulated by a physiotherapist while performing the proposed rehabilitation tasks.

All the developments (actuation system, control system and trajectory generation) have been experimentally validated on a test bench, enabling future trials on subjects to be envisaged. One of the prospects is to develop the control system for a second prototype actuated at the hip and knee, enabling repetition of the gait cycle.

This is the first prototype of its kind in the world.

Lower limb exoskeleton, rehabilitation exoskeleton, online trajectory generation, "assist-as-needed" command, actuation system.

Rapporteurs:
BABEL, Marie, Professeur des Universités, IRISA, INRIA Rennes - Bretagne Atlantique Campus Universitaire de Beaulieu.
MOHAMMED, Samer, Professeur des Universités, LISSI, Université de Paris-Est Créteil.

Examiners:
BOURHIS, Guy, Professeur des Universités, LCOMS, Université de Lorraine.

Examiners

Thesis supervisor:
PUDLO, Philippe. Professeur des Universités, LAMIH, CNRS, UMR 8201, Université Polytechnique Hauts-de-France.

Dissertation supervisor:
PUDLO, Philippe.

Co-supervisors:
DEQUIDT, Antoine. Maitre de Conférences, LAMIH, CNRS, UMR 8201, INSA Hauts-de-France.
BONNET, Xavier. Senior Lecturer, IBHGC, Arts et Métiers Paristech.

Guest:
DEHAIL, Patrick. Professeur des Universités - Praticien Hospitalier, Centre Hospitalier Universitaire de Bordeaux
.