An Observer for Estimating Translational Velocity from Optic Flow and Radar

Abstract

This thesis presents the development of a discrete time observer for estimating state information from optic flow and radar measurements. It is shown that estimates of translational and rotational speed can be extracted using a least squares inversion for wide fields of view or, with the addition of a Kalman Filter, for small fields of view. The approach is demonstrated in a simulated three dimensional urban environment on an autonomous quadrotor micro-air-vehicle (MAV). A state feedback control scheme is designed, whereby the gains are found via static <italic>H_&infin;</italic>, and implemented to allow trajectory following. The proposed state estimation scheme and feedback method are shown to be sufficient for enabling autonomous navigation of an MAV. The resulting methodology has the advantages of computational speed and simplicity, both of which are imperative for implementation on MAVs due to stringent size, weight, and power requirements.