Thesis defense of Amel Ait Ghezala-Sadoudi

The manual wheelchair (MRF) is commonly used by people with reduced mobility, but its daily use can lead to musculoskeletal disorders (MSDs) due to repetitive propulsion movements. In addition, accessibility constraints in urban environments often limit users' mobility, despite legislative advances in inclusion. Although electric wheelchairs (ERW) reduce physical effort, they have disadvantages such as limited autonomy and reduced physical activity, which can be detrimental in the long term. The manual wheelchair with electric assistance (FRM-AE) appears to be an intermediate solution, as it provides assistance while maintaining physical activity of the upper limbs. However, current assistance systems do not adapt effectively to users' abilities or to the constraints of the terrain.

The main challenge lies in designing an assistance system capable of taking into account the interaction between the user, the FRM-AE and the ground, offering precise and safe control. Another major challenge is the real-time validation of the system's functionality and robustness across a range of scenarios and topographies. User tests with a sufficiently large cohort are essential to assess the effectiveness and acceptability of these systems. However, testing in an outdoor environment is time-consuming and potentially risky. The use of an FRM simulator presents itself as an effective solution for overcoming these obstacles.

This thesis is divided into two main parts. The first part focuses on the development of a haptic and dynamic interface integrated into an FRM simulator, designed to simulate various locomotion scenarios. This interface incorporates a dynamic model of the interaction between the user, the FRM and the ground, as well as a haptic controller. The model can be used to reproduce movements such as walking in a straight line, turning, negotiating slopes and inclines. The haptic controller, based on a Linear Variable Parameter (LPV) optimal control approach, tracks the trajectories of the reference model while generating haptic resistances adapted to each movement. The interface also transmits kinematic and dynamic data to a hexapod platform, providing real-time sensory feedback. This system has been validated by tests on 30 able-bodied individuals and 2 people with reduced mobility, leading to the integration of an assistance system that takes into account the constraints of the terrain and the capabilities of the users.

The second part of the thesis focuses on the design of an assistance system centered on user capabilities and terrain constraints. Three assistance strategies have been developed. The first quantifies an accessibility difficulty index and integrates it into an assistance level parameter to modulate motor torque. The second integrates this parameter into an optimization algorithm via an adaptive cost function, enabling control to be adjusted according to the user's actions and the constraints of the terrain. The third strategy is based on adaptive assistance with an optimal assistance level parameter, enabling real-time adjustment of effort according to the user's needs. These three approaches were evaluated objectively and subjectively with 13 participants to compare their respective performances. Experimental validations, carried out using the FRM PSCHITT-PMR-LAMIH dynamic simulator, demonstrated the effectiveness of the solutions developed, particularly in terms of improved safety and comfort, thus validating the proposed strategies.

Reporters

• Mr. Lounis ADOUANE - Professor, Université de technologie de Compiègne
• Mr Rodolfo ORJUELA - Professor, Université de Haute-Alsace

Examiners

• Mrs. Marie BABEL - Professor, INSA de Rennes
• Mr. Yann MORERE - Maître de conférences HDR, Université de Lorraine

Invited

• Mr. Joseph BASCOU - Head of CERAH, CERAH

Thesis supervisors:

• Mr Chouki SENTOUH - Maître de conférences HDR, UPHF-INSA Hauts-de-France
• Mr Philippe PUDLO - Professeur, UPHF

Haptic interface, robust optimal control, real-time control, manual wheelchair simulator, adaptive assistance, user-centered assistance, User-FRM-ground interaction model, user model, online parametric estimation, user testing.