An Ultrasound Body Area Network for Arm Position Tracking and Arm Gesture Recognition

This thesis presents the concept and implementation of an ultrasound body area network for arm position tracking with potential application in the field of physio-therapeutic rehabilitation, and for arm gesture recognition in control applications. The network consists of multiple nodes that are equipped with ultrasound speakers and microphones and are mounted on a belt and both arms. Based on the transmitted and received ultrasound signals, it is possible to gain information on inter node distances, angles and relative speeds. This data is used to estimate the arm node positions with a maximum likelihood estimator and Kalman filters respectively. The resulting trajectory is applied to a gesture recognition algorithm based on Fourier descriptors. The three main tasks which are the acquisition of inter node information, the arm position estimation and gesture recognition are successfully implemented and tested on a demonstrator system.