Committee:

Dr. David Anderson, ECE, Chair , Advisor

Dr. Mark Davenport, ECE

Dr. Omer Inan, ECE

Dr. Benjamin Klein, ECE

Dr. Joshua Kacher, MSE

Abstract:

Prior Work in the GTBionics Lab at Georgia Tech has lead to the design of an assistive technology device for people with quadriplegia in the form of a wearable headset incorporating multiple magnetic sensors for measuring magnetic fields generated by a rare-earth magnet placed on a patient's tongue.  The aim is to track this magnet, and thereby the tongue's position, and then use those measurements to issue commands to a computer or smart phone.  Past versions of the headset have used classification techniques to accomplish this, but it has been a desideratum to provide more refined tracking and proportional control.  The intended nonlinear filtering algorithms require model estimation and model estimation requires access to calibrated measurements with known sensor orientations and positions.  The sensor positions are adjustable to the user, so it is necessary to calculate these parameters without the use of specialized equipment. The presented research describes a sequence of methods to synthesize the sensor measurements and estimate the model parameters; implementation considerations and potential problems are also presented.