Improving Real World Performance of Vision Aided Navigation in a Flight Environment

Abstract

The motivation of this research is to fuse information from an airborne imaging sensor with information extracted from satellite imagery in order to provide accurate position when GPS is unavailable for an extended duration. A corpus of existing geo-referenced satellite imagery is used to create a key point database. A novel algorithm for recovering coarse pose using by comparing key points extracted from the airborne imagery to the reference database is developed. This coarse position is used to bootstrap a local-area geo-registration algorithm, which provides GPS-level position estimates. This research derives optimizations for existing local-area methods for operation in flight environments.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 15, 2016
Accession Number
AD1017891

Entities

People

  • Donald T. Venable

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Air Platforms
  • Space

DTIC Thesaurus Topics

  • Air Force
  • Change Detection
  • Computational Science
  • Computer Vision
  • Databases
  • Detectors
  • Global Positioning Systems
  • Grids
  • Inertial Navigation
  • Inertial Navigation Systems
  • Kalman Filters
  • Mathematical Filters
  • Navigation
  • Simultaneous Localization And Mapping
  • Three Dimensional
  • Two Dimensional
  • World Geodetic System

Fields of Study

  • Computer science

Readers

  • Computer Vision.
  • Positioning, Navigation, and Timing (PNT) Technology.

Technology Areas

  • AI & ML
  • Space