| In this dissertation, a recursive Bayesian approach to kinematic registration using magnetic dipoles is proposed and an algorithm based on the Unscented Kalman filter (UKF) is formulated. The proposed model has applications in vehicle guidance in Intelligent Transportation Systems (ITS) and robot manipulator control. The general formulation of the components of the magnetic field equations for a given magnetic dipoles distribution is developed. This formulation does not impose any restriction on the magnetic dipole distributions. Then, it is shown that due to measuring device inaccuracy, the earth's magnetic field, the interactions of the magnetic fields and surrounding metallic objects, mean and variance does not fully capture the statistical properties of the magnetic field components in presence of noise. Using Monte Carlo simulations, the ability of the Unscented Transformation in capturing higher moments of the magnetic dipole functions is investigated.;A general continuous stochastic model capturing the kinematics of a rigid body moving in a known vector magnetic field is established. A discrete 21-state UKF design based on the proposed stochastic dynamical model is formulated. Simulation results for Robot manipulator control and vehicle guidance in ITS indicate that the algorithm is robust and performs effectively.;The overall aim of this work is to derive a computationally efficient algorithm for using magnetic dipoles to optimally estimate states of nonlinear stochastic dynamical systems in presence of uncertainty. |
