Theses and Dissertations

Jacob Bowen

Issuing Body

Mississippi State University

Advisor

Jones, Bryan A

Committee Member

Morris, Thomas H

Committee Member

Dandass, Yoginder

Date of Degree

12-15-2012

Document Type

Graduate Thesis - Open Access

Major

Computer Engineering

Degree Name

Master of Science

College

James Worth Bagley College of Engineering

Department

Department of Electrical and Computer Engineering

Abstract

Micro Aerial Vehicles (MAVs) provide a highly capable, agile platform, ideally suited for intelligence/surveillance/reconnaissance missions, urban search and rescue, and scientific exploration. Critical to the success of these tasks is a system which moves au-tonomously through an unknown, obstacle-strewn, GPS-denied environment. Classical simultaneous localization and mapping (SLAM) approaches rely on large, heavy sensors to generate 3-D information about a MAV’s surroundings, severely limiting its abilities. This motivates a study of Parallel Tracking and Mapping (PTAM), an algorithm requiring only a single camera to provide 3-D data to an autonomous navigation system. Metric properties of 3-D MAV pose estimates are compared with physical measurements to ex-plore tracking accuracy. Additionally, a discrete wavelet transform-based keypoint detec-tor is implemented for a feasibility study on improving map density in low-visual-detail environments. Finally, a system is presented that integrates PTAM, autonomous MAV control, and a human interface for manual control and data logging.

URI

https://hdl.handle.net/11668/20322

Comments

UAV||MAV||SLAM||Autonomous Navigation||Computer Vision||Keypoint Detection

Download

Share

COinS