Motion Planning for Semi-Morphing Unmanned Aerial Vehicles Operating in Confined Environments

Abstract

In recent decades, there has been a notable surge in scholarly interest focused on Unmanned Aerial Vehicles (UAVs). The development of advanced UAVs having the abilities to per- form complex flight maneuvers, fly inside complex spaces, and execute intricate missions in changing environments is garnering considerable attention. These environments include sectors such as mining, Urban Search and Rescue (USAR), military operations, and a range of industrial applications including the maintenance and repair of subterranean infrastructures. The pressing demand for accessing and operating within confined spaces has become a driving force compelling researchers to advance UAV technologies. These advancements are aimed at overcoming the complexities associated with operating in constrained environments and addressing the current limitations of UAVs, while concurrently enhancing their overall performance capabilities. In this thesis, a set interconnected tools targeted to enable UAVs to autonomously plan their flight maneuvers inside confined spaces is presented. To achieve this, an Improved “Teach – Repeat – Replan” (I-TRP) iterative strategy is formulated. The solution is a hybrid offline-online approach encompassing four major modules across a three-phases strategy. An advanced 3D flight corridor with a novel Occupancy Checking property is developed based on a hand-crafted path (Teaching phase) and the perceived environmental geometrical features. Additionally, an enhanced version of a common global path planning algorithm, Field D*, is developed in combination with the generated flight corridor to formulate a prior near-optimal and smooth topological equivalent path through an offline process (Repeating phase). Finally, a local planning algorithm with online collision checking and obstacle avoidance is formulated via a Sequential Convex Optimization process (Replanning phase). Such local plan is utilized to generate a posterior optimized dynamically feasible path using the terrain information as captured by the UAV’s on-board sensors. The posterior reference path is used to formulate a set of guidance commands comprising the position, attitude, velocity, and acceleration of the aircraft to guide the UAV’s flight within the generated (potentially having complex geometrical features) flight corridor. The developed path following methodology is formulated via the use of a Non-linear Model Pre- dictive formulation. The developed I-TRP strategy guides autonomous UAVs to reach their destination within practically any structured or unstructured environment presenting diverse degrees of geometrical complexity ranging between open free spaces to highly cluttered environments. Simulation results showcase the capabilities of the developed I-TRP strategy outperform- ing available mechanisms in a computational efficient process suitable for real time flight navigation.