Design, implementation and experimentation of an assistive control strategy for lower-limb underactuated exoskeletons

Ageing of society and the consequent need to develop assistive devices able to improve the quality of daily life in elderly people are the key issues of this work. The final objective of this work is the evaluation of the feasibility of an under-actuated lower-limb wearable orthosis. The starting point of this thesis was the identification of physiological torque reference for the hip and knee joints during different activities of daily living, and a subsequent development of suitable torque profiles for driving hip and knee actuated joints. Then, a control architecture was developed for the control of an Active Pelvis Orthosis (APO) and a Knee-Ankle-Foot-Orthosis (KAFO). The control architecture is composed of a Phase Estimator and a Locomotion Mode Classifier. The Phase Estimator is an algorithm based on adaptive oscillators capable of learning from the user to predict its gait pattern, allowing for a reliable estimation of the walking phase at any time. This allows for a continuous reference for the assistive torque application. The Locomotion Mode Classifier is an algorithm used for quick and reliable identification of the user’s high level intention. Furthermore, the concept of synergies across the hip and knee joints was explored. In order to simplify the concept of the future under-actuated orthosis, hence hopefully reducing its weight and cost, hip and knee joints’ behavior were studied, looking for the existence of synergies between the moment profiles. If such similarities should prove feasible, it would be possible to simplify the future transmission system of the under-actuated orthosis