In this study, we present a tomography-based control system for a rehabilitation robot using a novel approach to assess advancement and a dynamic model of the system. In this model, the torque generated by the robot and the impedance of the patient's hand are used to determine each step of the rehabilitation. In the proposed control architecture, a regression model is developed and implemented based on the extraction of tomography signals to estimate the muscles state. During the rehabilitation session, the torque applied by the patient is adjusted according to this estimation. The first step of this protocol is to calculate the subject-specific parameters. These include the axis offset, inertia parameters, passive damping and stiffness. The second step involves identifying the other elements of the model, such as the torque resulting from interaction. In this case, the robot will calculate the torque generated by the patient. The developed robot-based solution and the suggested protocol were tested on different participants and showed promising results. First, the prediction of the impedance-position relationship was evaluated, and the prediction was below 2% error. Then, different participants with different impedances were tested, and the results showed that the control system controlled the force and position for each participant individually.
Keywords: artificial intelligence; control; home rehabilitation; impedance tomography.