Range-Based Localization of UAVs with Moving Ultra-Wideband Sensors

Abstract

This thesis focuses on the distance-based localization of Unmanned Aerial Vehicles(UAVs) using Ultra-Wideband(UWB) sensors and moving beacons with rigid formation. In distance-based localization problems, gradient based localization laws have been dominantly used. Using the conventional gradient descent law in the rigid beacon settings however will cause the total error to increase as the agent moves farther from the beacons, leading to instability and slow convergence issues. We propose an adaptive localization law that utilizes RMS Prop technique to change the gradient descent gains adaptive to the distance between mobile node and beacons, thus giving the improved convergence speed and provide stable localization. Among various optimization methods such as ADAgrad and Adadelta, RMS Prop is chosen because of its simple utilization and analysis. The comparison of a Modified Least-Squares(MLS) based algorithm and the proposed RMS Prop based algorithm is provided to analyze the advantages and disadvantages over simulated environment. Then we implement four rigid moving beacons installed on UAVs to provide consistent local coordinate information regardless of the set-up environment. The moving beacon scheme has the advantages of removing the inconvenient beacon installation process and disconnection possibilities. By using this dynamic with the previous RMS Prop method, the tag agent can localize the local coordinate position as the beacon agents moves in translational motion, which confirms the feasibility of moving beacon implementation.