Abstract

We demonstrate a multiple micro aerial vehicle (MAV) system capable of supporting autonomous exploration and navigation in unknown environments using only a sensor commonly found in low-cost, commercially available MAVs—a front-facing monocular camera. We adapt a popular open source monocular SLAM library, ORB-SLAM, to support multiple inputs and present a system capable of effective cross-map alignment that can be theoretically generalized for use with other monocular SLAM libraries. Using our system, a single central ground control station is capable of supporting up to five MAVs simultaneously without a loss in mapping quality as compared to single-MAV ORB-SLAM. We conduct testing using both benchmark datasets and real-world trials to demonstrate the capability and real-time effectiveness.

Library of Congress Subject Headings

Drone aircraft--Control systems; Robots--Control systems; Micro air vehicles--Control systems; Remote sensing

Publication Date

8-2018

Document Type

Thesis

Student Type

Graduate

Degree Name

Computer Science (MS)

Department, Program, or Center

Computer Science (GCCIS)

Advisor

Zack Butler

Advisor/Committee Member

Minseok Kwon

Advisor/Committee Member

Joseph Geigel

Campus

RIT – Main Campus

Plan Codes

COMPSCI-MS

Download

Share

COinS