Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhanced Monocular Visual Odometry with AR Poses and Integrated INS-GPS for Robust Localization in Urban Environments

Ankit Shaw

TL;DR

This paper introduces a cost effective localization system combining monocular visual odometry, augmented reality (AR) poses, and integrated INS-GPS data, and enhances accuracy with INS and GPS data, filtered through an Extended Kalman Filter.

Abstract

This paper introduces a cost effective localization system combining monocular visual odometry , augmented reality (AR) poses, and integrated INS-GPS data. We address monocular VO scale factor issues using AR poses and enhance accuracy with INS and GPS data, filtered through an Extended Kalman Filter . Our approach, tested using manually annotated trajectories from Google Street View, achieves an RMSE of 1.529 meters over a 1 km track. Future work will focus on real-time mobile implementation and further integration of visual-inertial odometry for robust localization. This method offers lane-level accuracy with minimal hardware, making advanced navigation more accessible.

Enhanced Monocular Visual Odometry with AR Poses and Integrated INS-GPS for Robust Localization in Urban Environments

TL;DR

This paper introduces a cost effective localization system combining monocular visual odometry, augmented reality (AR) poses, and integrated INS-GPS data, and enhances accuracy with INS and GPS data, filtered through an Extended Kalman Filter.

Abstract

This paper introduces a cost effective localization system combining monocular visual odometry , augmented reality (AR) poses, and integrated INS-GPS data. We address monocular VO scale factor issues using AR poses and enhance accuracy with INS and GPS data, filtered through an Extended Kalman Filter . Our approach, tested using manually annotated trajectories from Google Street View, achieves an RMSE of 1.529 meters over a 1 km track. Future work will focus on real-time mobile implementation and further integration of visual-inertial odometry for robust localization. This method offers lane-level accuracy with minimal hardware, making advanced navigation more accessible.

Paper Structure

This paper contains 15 sections, 11 equations, 11 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Method
  4. Google Street View Data
  5. Feature Matching
  6. Visual Pose Estimation
  7. Fixed-pose Local Bundle Adjustment
  8. Standard Local Bundle Adjustment
  9. Sensor Fusion
  10. Visual Odometry and AR Poses
  11. INS and Estimated GPS
  12. Testing Methodology
  13. Results
  14. Future Work
  15. Conclusion

Figures (11)

  • Figure 1: Diagram illustrating my system pipeline
  • Figure 2: Google Street View Panorama (RGB and Depth)
  • Figure 3: From left to right: captured frame and nearby panorama image, camera frame and panorama SIFT key points, camera frame and panorama FLANN matches.
  • Figure 4: Visualisation of a Virtual Line Descriptor kvld
  • Figure 5: Top: Camera frame and panorama with consistent VLDs highlighted. Bottom: K-VLD filtered matches
  • ...and 6 more figures