A model based stroke rehabilitation monitoring system using inertial navigation systems

In this thesis, a low-cost rehabilitation monitoring system based on Inertial Navigation System (INS) techniques is developed. The purpose of this system is to provide practitioners with data sufficient-enough to characterise a post-stroke patient’s arm motion. We propose an inertial sensor-based system to characterize the motion trajectory. The system is composed of two main components: joint angle estimation and local positioning of points of interest. Joint angles are estimated using an Extended Kalman Filter by fusing data from different sensors. Local positioning is done using forward kinematic solutions: homogenous transformation, Denavit Hartenberg’s method and dual quaternions that approximating the arm as a serial robot. As a result, the relative joint angles and 3D positioning of points of interests on the human are calculated accurately. The data obtained prove useful for reconstructing the arm’s three-dimensional motion trajectories and provide feedback to stroke survivor patients or (remotely) the rehabilitation practitioner.