Wearable sensor based gesture recognition system

Meroni, Charlotte (2016) Wearable sensor based gesture recognition system. Bachelor thesis, Scuola Universitaria professionale della Svizzera italiana (SUPSI).

Full text not available from this repository.

Abstract

As part of the Internet of Things, we want to develop a wearable device that allows to recognize gestures made with hand and arm. With the Shimmer device, which has sensors able to record the forearm inertial data and the signals generated by the forearm muscles (EMG), we want to study the characteristics of the acquired signals in order to define the right features that can be used for classifying the movements. In this way it is possible to use the limb as a natural interface for controlling complex instruments and systems. After the learning phase of the functionalities of the Shimmer devices and the characteristics and peculiarities of the electromyographic signal, an application capable of recognizing specific hands movements has been developed. The results obtained from the application tests show an accuracy of 82.8% with a simple tree classifier, which increases dramatically up to 98.2% with a linear SVM classifier. The application has been developed in a way that other movements can be added to the first set, at the only conditions that an appropriate number of acquisitions is recorded and the classifier is trained again. The key point of this project is the acquisition and analysis of the electromyographic signal; indeed, this study can be applied to the medical field where the prosthetic biomechanical implants are interfaced use muscle impulses in order to move the artificial limb like a real one.

Item Type: Thesis (Bachelor)
Subjects: Informatica
Divisions: Dipartimento tecnologie innovative > Ingegneria informatica
URI: http://tesi.supsi.ch/id/eprint/1033

Actions (login required)

View Item View Item