Date of Award

3-2021

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Electrical and Computer Engineering

First Advisor

Scott L. Nykl, PhD

Abstract

The goal of AAR is to control the tanker boom to safely refuel a receiving aircraft with no input or aid from the boom operator. To achieve this, the pose of the receiver relative to the tanker must be known. Point set registration is a fundamental issue used to estimate the relative pose of an object in an environment. However, it's likely a computational bottleneck of a vision processing pipeline. In addition, the matching of each sensed point with a closest truth point, nearest neighbor matching, is the most costly portion of the point set registration process. For this reason, this research focuses on speeding up the ICP algorithm and nearest neighbor algorithms. This research lays out novel nearest neighbor matching algorithms based on the Delaunay triangulation with a reduced cost compared to conventional nearest neighbor matching algorithms. The ICP algorithm is transformed into a massively parallel algorithm and mapped onto a vector processor to realize a speedup of approximately 2 orders of magnitude. Lastly, this thesis presents algorithmic and runtime analysis with augmented, virtual, and real experiments.

AFIT Designator

AFIT-ENG-MS-21-M-075

DTIC Accession Number

AD1134593

Download

Share

COinS