Intelligent Unmanned Aerial Vehicles for Inspecting Indoor Complex Aero-Structures

Abstract

In recent years there has been an increasing need to monitor and inspect the integrity of structures that are usually large and geometrically complex. In an attempt to facilitate this type of activity, mapping and 3D constructing are performed on these structures to generate reference maps with high resolution. These structures can be inspected with unmanned aerial vehicles (UAVs) due to their ability to move freely in space, but these vehicles have limited ight duration. In this thesis, a method is proposed to reduce flight time and ensure coverage completeness by performing a rough initial scan of the structure. This rough map is then used to guide a more thorough Next-Best-View (NBV) approach to creating a dense reconstruction of the object that exploits reectional symmetry in the structure. We have successfully implemented an NBV approach with a novel utility function which combines information theory, distance, model density and predictive measures based on symmetries in the structure. The justification for the technique is provided, along with experimental results. This system outperforms classic information gain approaches with a shorter ight path and higher coverage completeness.

Date of Award	Jun 2017
Original language	American English
Supervisor	Dongming Gan (Supervisor)