MOTION PLANNING FOR MULTI-ROTOR UAVS IN CLUTTERED ENVIRONMENT

Abstract

Multi-rotor Unmanned Aerial Vehicles (UAVs) have been an increasingly popular platform for research, commercial and other applications. In this thesis, we will examine motion planning algorithms for UAVs as one of the fundamental components to support autonomous navigation. Leveraging on State-Of-The-Arts algorithms, we propose a framework that generates dynamically feasible, collision-free, time-parameterized trajectory to support UAV's autonomous mission in cluttered environment. We first introduce path-searching methods that find a collision-free geometric path in the given planning environment. The result is used as a guiding path to generate Safe Flight Corridors (SFC) that represents the traversable space. Collision-free waypoints are then extracted from the SFCs and used as initial guess for the succeeding trajectory optimization module which minimizes control effort, trajectory duration and satisfies various user-defined safety requirements. We emphasize the practicality of the proposed framework with preliminary verification in MATLAB, implementation in C++ and benchmark testing using Robot Operating System.