Development of techniques and algorithms for the functional evaluation and assistance in the movements of the upper limb

Abstract

This thesis describes some novel techniques and algorithms for the functional evaluation and assistance in the movements of the upper limb. In this context and with particular focus on Parkinson’s disease, the framework of the development of innovative solutions is here described and new proposals are presented. Different methods for gathering and exploiting information measured from the patients are deeply investigated and described in this thesis. The main theories and computational models of the motor control of the upper limb are here described and classified depending on their characteristics, in terms of biologically inspired design. This classification helped getting a better understanding of the upper limb control mechanisms for the intelligent design of complementary control systems for assistive devices. The principal findings and some novel approaches in the development of devices for tremor suppression in the upper limb are described and this knowledge is used for the development of a original neural-based control architecture able to limit tremor in the arm using Functional Electrical Stimulation (FES). The importance of models as benchmark platform for the development of assistive devices is highlighted and the principal steps for the development of a biomechanical 3D model of the upper limb as simulative tool for the design of tremor intervention strategies are addressed. Some novel algorithms for the extraction of information on movements are developed and presented. Information about movement timing is extracted in real-time through the analysis of the surface electromyographic signal (sEMG), independently from the level of the noise affecting the signal. Algorithms for the classification in real-time between voluntary and involuntary movements from electroencephalographic and movement-related signal are described and applied to signals extracted from healthy and impaired persons. Finally, a novel approach on the assessment of muscular rigidity through the analysis of voice signals is presented, in the framework of the design of wearable sensors-based solutions for the evaluation of the symptoms of patients affected by Parkinson’s disease.