Pilot Study of a Performance-Based Adaptive Assistance Controller for Stroke Survivors

Robotic gait training is a promising tool to improve walking ability after neurological disorders. Despite its benefits, its therapeutic effects might depend on the customization of the robotic training to the patient's capabilities. In the last years, various approaches focused on the assist-as-needed (AAN) paradigm, have been proposed to improve the effectiveness of robotic rehabilitation. However, in most of these cases the amount of robotic assistance is chosen by a therapist, and therefore, subjective decisions could affect the adaptation of the therapy to the patient's needs. This contribution presents the implementation of a novel performance-based adaptive controller, which automatically adjusts robotic assistance for diverse subtasks of gait based on the user's performance during training. A pilot study testing this controller in one stroke survivor, shows the potential of the proposed tool to be included in future robot-based gait training protocols.