DM-VIO: Delayed Marginalization Visual-Inertial Odometry

TL;DR

DM-VIO introduces delayed marginalization and pose graph bundle adjustment to enhance monocular visual-inertial odometry, enabling better IMU integration, scale estimation, and outperforming existing methods across diverse datasets.

Contribution

The paper proposes delayed marginalization and pose graph bundle adjustment techniques, allowing for improved IMU integration and more accurate scale estimation in visual-inertial odometry.

Findings

Outperforms state-of-the-art visual-inertial odometry methods.

Effectively captures full photometric uncertainty for better scale estimation.

Successfully handles diverse scenarios including drone, handheld, and automotive environments.

Abstract

We present DM-VIO, a monocular visual-inertial odometry system based on two novel techniques called delayed marginalization and pose graph bundle adjustment. DM-VIO performs photometric bundle adjustment with a dynamic weight for visual residuals. We adopt marginalization, which is a popular strategy to keep the update time constrained, but it cannot easily be reversed, and linearization points of connected variables have to be fixed. To overcome this we propose delayed marginalization: The idea is to maintain a second factor graph, where marginalization is delayed. This allows us to later readvance this delayed graph, yielding an updated marginalization prior with new and consistent linearization points. In addition, delayed marginalization enables us to inject IMU information into already marginalized states. This is the foundation of the proposed pose graph bundle adjustment, which we use for IMU initialization. In contrast to prior works on IMU initialization, it is able to capture the full photometric uncertainty, improving the scale estimation. In order to cope with initially unobservable scale, we continue to optimize scale and gravity direction in the main system after IMU initialization is complete. We evaluate our system on the EuRoC, TUM-VI, and 4Seasons datasets, which comprise flying drone, large-scale handheld, and automotive scenarios. Thanks to the proposed IMU initialization, our system exceeds the state of the art in visual-inertial odometry, even outperforming stereo-inertial methods while using only a single camera and IMU. The code will be published at http://vision.in.tum.de/dm-vio