Direkt zum Inhalt springen
Computer Vision Group
TUM School of Computation, Information and Technology
Technical University of Munich

Technical University of Munich

Menu

Links

Informatik IX
Computer Vision Group

Boltzmannstrasse 3
85748 Garching info@vision.in.tum.de

Follow us on:

News

04.03.2024

We have twelve papers accepted to CVPR 2024. Check our publication page for more details.

18.07.2023

We have four papers accepted to ICCV 2023. Check out our publication page for more details.

02.03.2023

CVPR 2023

We have six papers accepted to CVPR 2023. Check out our publication page for more details.

15.10.2022

NeurIPS 2022

We have two papers accepted to NeurIPS 2022. Check out our publication page for more details.

15.10.2022

WACV 2023

We have two papers accepted at WACV 2023. Check out our publication page for more details.

More


Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
research:vslam:vi-dso [2017/09/20 02:22]
Lukas von Stumberg Added supplementary material and video.
research:vslam:vi-dso [2018/04/17 17:22]
Lukas von Stumberg
Line 1: Line 1:
-==== Direct Sparse Visual-Inertial Odometry ==== +====== VI-DSO: Direct Sparse Visual-Inertial Odometry using Dynamic Marginalization ====== 
- +**Contact:** [[members:stumberg|Lukas von Stumberg]], [[members:usenko]], [[members:cremers|ProfDaniel Cremers]]
-This page provides supplementary material and a video to the paper "Direct Sparse Visual-Inertial Odometry using Dynamic Marginalization". +
- +
-The supplementary material can be downloaded at{{ :research:vslam:vi-dso:vi-dso-supplementary-material.pdf |}}. +
-The video is available at: [[https://youtu.be/5fsIvWg46oc|https://youtu.be/5fsIvWg46oc]]+
  
 <html> <html>
-<iframe width="560" height="315" src="https://www.youtube.com/embed/5fsIvWg46oc?rel=0" frameborder="0" allowfullscreen></iframe>+<iframe width="560" height="315" src="https://www.youtube.com/embed/GoqnXDS7jbA?rel=0" frameborder="0" allowfullscreen></iframe>
 </html> </html>
 +
 +===== Abstract =====
 +We present **VI-DSO**, a novel approach for visual-inertial odometry,
 +which jointly estimates camera poses and sparse scene
 +geometry by minimizing photometric and IMU measurement
 +errors in a combined energy functional. The visual part of
 +the system performs a bundle-adjustment like optimization
 +on a sparse set of points, but unlike key-point based
 +systems it directly minimizes a photometric error. This
 +makes it possible for the system to track not only corners,
 +but any pixels with large enough intensity gradients. IMU
 +information is accumulated between several frames using
 +measurement preintegration, and is inserted into the
 +optimization as an additional constraint between keyframes.
 +We explicitly include scale and gravity direction into our
 +model and jointly optimize them together with other
 +variables such as poses. As the scale is often not
 +immediately observable using IMU data this allows us to
 +initialize our visual-inertial system with an arbitrary
 +scale instead of having to delay the initialization until
 +everything is observable. We perform partial
 +marginalization of old variables so that updates can be
 +computed in a reasonable time. In order to keep the system
 +consistent we propose a novel strategy which we call
 +"dynamic marginalization". This technique allows us to use
 +partial marginalization even in cases where the initial
 +scale estimate is far from the optimum. We evaluate our
 +method on the challenging EuRoC dataset, showing that VI-DSO outperforms the state of the art.
 +
 +===== Downloads =====
 +The paper can be downloaded at: http://arxiv.org/abs/1804.05625 \\
 +There is also supplementary material with additional evaluation and mathematical derivations at: {{ :research:vslam:vi-dso:vi-dso-supplementary-material.pdf |}}. \\
 +The video is available at: [[https://youtu.be/GoqnXDS7jbA|https://youtu.be/GoqnXDS7jbA]]
 +
 +The project is based on [[:research:vslam:dso|DSO]] which was developed by Jakob Engel.
 +
 +<bibtex>
 +<keywords>vi-dso</keywords>
 +</bibtex>

Rechte Seite

Informatik IX
Computer Vision Group

Boltzmannstrasse 3
85748 Garching info@vision.in.tum.de

Follow us on:

News

04.03.2024

We have twelve papers accepted to CVPR 2024. Check our publication page for more details.

18.07.2023

We have four papers accepted to ICCV 2023. Check out our publication page for more details.

02.03.2023

CVPR 2023

We have six papers accepted to CVPR 2023. Check out our publication page for more details.

15.10.2022

NeurIPS 2022

We have two papers accepted to NeurIPS 2022. Check out our publication page for more details.

15.10.2022

WACV 2023

We have two papers accepted at WACV 2023. Check out our publication page for more details.

More