REMAID

Various activities are offered, such as physiotherapy, occupational therapy and adapted physical activity, but also approaches based on assistive technologies (e.g., robotic training or functional electrical stimulation). These assistive technologies are recommended by the French National Authority for Health as a complement to conventional treatments. However, the high cost of these assistive technologies limits their use in rehabilitation departments. LAMIH's SHV department has recently developed and patented a motorized ergometer with a reasonable production cost. This space-saving measurement and rehabilitation tool can be transported to the patient's bed, enabling early rehabilitation treatment. This rehabilitation tool has recently been coupled with a Brain Machine Interface (BMI) to mobilize altered joints based on brain activation alone. MCIs are technological tools for processing brain electrical activity (EEG) in order to identify specific activity such as, for example, brain activity associated with voluntary motor skills. The main idea is to be able to discriminate motor intention, i.e., before the motor action is actually performed. In order to improve the effectiveness of rehabilitative MCI, a great deal of work has been carried out to discriminate the motor parameters of the forthcoming movement (e.g., force, speed of execution and direction of the forthcoming movement) directly from EEG signals. This work seeks to develop a hybrid ICM coupling electrophysiological (EEG, EMG) and biomechanical (force moment, kinematics) data in order to generate a personalized rehabilitative movement. The motor intention will be detected from the EEG signals to initiate the movement in the right temporal timing. During the execution phase, the movement will be corrected using EMG control signals and force moment. From a technical point of view, these constraints will be resolved by developing fast control laws that can act in a predictive mode from a partially identified setpoint, with possible reconfiguration as required. Depending on the complexity of the model used, it is also planned to use decomposition methods or tools derived from non-linear systems such as Takagi-Sugeno models. The dual aim of this work will therefore be  to validate the best algorithms for extracting instructions from electrophysiological and biomechanical signals, and to build the control laws for generating the personalized rehabilitative movement.