Cooperative maneuver enabled UAV relative localization.

摘要

The ability to localize itself in an environment is essential for any autonomous vehicle. In applications involving multiple Unmanned Aerial Vehicles (UAVs) such as formation flight, surveillance, and mapping, the relative pose of each UAV provides vital information for multiple vehicles to coordinate effectively. The majority of cooperative localization tasks involving multiple UAVs assume the transformation between reference frames, but this information is not always provided in the absence of landmarks or a Global Positioning System (GPS).;This thesis presents a method for estimating the relative pose of a pair of UAVs using range-only measurements. In this method, there is no prior information assumed about the relative pose of each UAV. Instead, the trajectories are constrained in order to leverage the circumstances in which the range-only measurements are obtained during flight. To augment this limited information, motion is used to construct a graph of the UAV trajectories using only distance measurements. Using the constructed graph, four potential solutions are derived for the relative pose, and a batch estimate is performed to obtain numerical estimates of each of the four solutions. Using the batch estimate, a cooperative maneuver can be performed in order to obtain a unique solution. The sensitivity to the trajectory and measurement noise are analyzed through a Monte Carlo analysis.