Design and implementation of a training strategy in chronic stroke with an arm robotic exoskeleton

Abstract

The distinguishing features of active exoskeletons are the capability of guiding arm movement at the level of the full kinematic chain of the human arm, and training full 3D spatial movements. We have specifically developed a PD sliding mode control for upper limb rehabilitation with gain scheduling for providing "assistance as needed", according to the force capability of the patient, and an automatic measurement of the impaired arm joint torques, to evaluate the hypertonia associated to the movement during the execution of the training exercise. Two different training tasks in Virtual Reality were devised, that make use of the above control, and allow to make a performance based evaluation of patient's motor status. The PERCRO L-Exos (Light-Exoskeleton) was used to evaluate the proposed algorithms and training exercises in two clinical case studies of patients with chronic stroke, that performed 6 weeks of robotic assisted training. Clinical evaluation (Fugl-Meyer Scale, Modified Ashworth Scale, Bimanual Activity Test) was conducted before and after treatment and compared to the scores and the quantitative indices, such as task time, position/joint error and resistance torques, associated to the training exercises.