Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Improving Monocular Visual-Inertial Initialization with Structureless Visual-Inertial Bundle Adjustment

Junlin Song, Antoine Richard, Miguel Olivares-Mendez

TL;DR

This work targets monocular visual-inertial initialization under dynamic motion by refining a structureless initialization with a novel structureless VI-BA. The method replaces 3D landmark-based constraints with epipolar geometry while preserving IMU preintegration, enabling accurate, real-time refinement of initial VIO states without 3D map reconstruction. Experimental results on EuRoC and TUM-VI show substantial improvements in translation and rotation accuracy and velocity estimates over strong baselines, with competitive compute times. The approach offers a practical, efficient path to robust VIO initialization, and invites future exploration of multi-view structureless constraints.

Abstract

Monocular visual inertial odometry (VIO) has facilitated a wide range of real-time motion tracking applications, thanks to the small size of the sensor suite and low power consumption. To successfully bootstrap VIO algorithms, the initialization module is extremely important. Most initialization methods rely on the reconstruction of 3D visual point clouds. These methods suffer from high computational cost as state vector contains both motion states and 3D feature points. To address this issue, some researchers recently proposed a structureless initialization method, which can solve the initial state without recovering 3D structure. However, this method potentially compromises performance due to the decoupled estimation of rotation and translation, as well as linear constraints. To improve its accuracy, we propose novel structureless visual-inertial bundle adjustment to further refine previous structureless solution. Extensive experiments on real-world datasets show our method significantly improves the VIO initialization accuracy, while maintaining real-time performance.

Improving Monocular Visual-Inertial Initialization with Structureless Visual-Inertial Bundle Adjustment

TL;DR

This work targets monocular visual-inertial initialization under dynamic motion by refining a structureless initialization with a novel structureless VI-BA. The method replaces 3D landmark-based constraints with epipolar geometry while preserving IMU preintegration, enabling accurate, real-time refinement of initial VIO states without 3D map reconstruction. Experimental results on EuRoC and TUM-VI show substantial improvements in translation and rotation accuracy and velocity estimates over strong baselines, with competitive compute times. The approach offers a practical, efficient path to robust VIO initialization, and invites future exploration of multi-view structureless constraints.

Abstract

Monocular visual inertial odometry (VIO) has facilitated a wide range of real-time motion tracking applications, thanks to the small size of the sensor suite and low power consumption. To successfully bootstrap VIO algorithms, the initialization module is extremely important. Most initialization methods rely on the reconstruction of 3D visual point clouds. These methods suffer from high computational cost as state vector contains both motion states and 3D feature points. To address this issue, some researchers recently proposed a structureless initialization method, which can solve the initial state without recovering 3D structure. However, this method potentially compromises performance due to the decoupled estimation of rotation and translation, as well as linear constraints. To improve its accuracy, we propose novel structureless visual-inertial bundle adjustment to further refine previous structureless solution. Extensive experiments on real-world datasets show our method significantly improves the VIO initialization accuracy, while maintaining real-time performance.

Paper Structure

This paper contains 13 sections, 20 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. Notation
  4. Proposed Initialization Framework
  5. Visual-Inertial Bundle Adjustment
  6. Structure-based VI-BA
  7. IMU preintegration residual
  8. Visual reprojection residual
  9. Structureless VI-BA
  10. Results
  11. EuRoC dataset
  12. TUM-VI dataset
  13. Conclusion

Figures (4)

  • Figure 1: Structureless VIO Initialization framework enhanced by structureless VI-BA (highlighted in blue).
  • Figure 2: Left: factor graph for structure-based VI-BA. Right: factor graph for structureless VI-BA.
  • Figure 3: Co-planar geometric relationships for feature bearing vectors with the frame-to-frame translation vector.
  • Figure 4: Angular and linear velocity profiles of MH_01_easy from EuRoC dataset and room1 from TUM-VI dataset.