Robot-aided assessment of walking function based on an adaptive algorithm

Abstract

Valid, sensitive and reliable assessments of walking recovery are needed to monitor the progresses of the patients throughout the whole rehabilitation course. Although instrumented assessment methods of walking and walking-related functions exist, they are seldom used in the clinical practice due to the long time required to administer them. The long-term goal of this project is to develop real-time assessment algorithm of walking ability that can be integrated in therapy with a robotic gait trainer. The assessment method is based on an adaptive controller that modifies the level of support as provided by i) the mechanical impedance of the robot's hip and knee joints and ii) the body weight support provided by the device. The level of support is adapted according to the subjects' ability to maintain a physiological walking pattern. We hypothesize that the required level of support is proportional to the walking impairment of the patient. A measure of the robotic assistance necessary for walking in a robotic gait trainer can be used as an assessment of the residual capacity of the subject. We present a first experimental evaluation of the assessment algorithm in 8 healthy subjects using the Lokomat (Hocoma AG, Switzerland), a commercially-available robotic gait trainer. Our results suggest that this method can be a promising tool to objectively assess walking function during training in the clinical practice.