An adaptive assistance algorithm for robot-based neurorehabilitation therapy of the human upper limb

This thesis proposes an algorithm for real-time robotic assistance tuning, which could be exploited in robot-based therapy with any kind of active device for upper-limb neurorehabilitation. After a general overview on rehabilitation and robotic devices for health care, this study illustrates in details the following: (i) the serious game implemented to present the rehabilitation task in a virtual environment, (ii) the method to evaluate patient's performance, and (iii) the actual algorithm for robotic assistance tuning. Therefore, the work describes in detail how to extract accurate performance indices from the subject’s execution, and how to correlate them with the amount of assistance to be provided. The algorithm also aims at enhancing subject’s efforts for a more effective recovery, tailoring the therapy to the patient without prior knowledge of his/her physical situation. Finally, several assessment phases also illustrate the effectiveness of the procedure, showing how the system adapts to the difficulties experienced by the subjects and whether the system is effective for rehabilitation of post-stroke patients.