Direkt zum Inhalt springen
Computer Vision Group
TUM Department of Informatics
Technical University of Munich

Technical University of Munich

Menu
Home Research Areas Visual SLAM GN-Net: The Gauss-Newton Loss for Deep Direct SLAM

GN-Net: The Gauss-Newton Loss for Deep Direct SLAM

Abstract

Direct methods for SLAM have shown exceptional performance on odometry tasks. However, they still suffer from dynamic lighting/weather changes and from a bad initialization on large baselines. To mitigate both of these effects, we propose an approach which feeds deep visual descriptors for each pixel as input to the SLAM system. In this work, we introduce GN-Net: a network optimized with the novel Gauss-Newton loss for training deep features. It is designed to maximize the probability of the correct pixel correspondence inside the Gauss-Newton algorithm. This results in features with a larger convergence basin when compared with single-channel grayscale images generally used in SLAM-based approaches. Our network can be trained with ground-truth pixel correspondences between different images, produced either from simulation data or by any state-of-the-art SLAM algorithm. We show that our approach is more robust against bad initialization, variations in day-time, and weather changes thereby outperforming state-of-the-art direct and indirect methods. Furthermore, we release an evaluation benchmark for what we refer to as relocalization tracking. It has been created using the CARLA simulator as well as sequences taken from the Oxford RobotCar Dataset.

Downloads

The paper can be downloaded at: http://arxiv.org/abs/1904.11932
The video is available at: https://youtu.be/GleEp0PqAv0

The dataset will be published soon.


Export as PDF, TEX or BIB

Preprints
2019
[]GN-Net: The Gauss-Newton Loss for Deep Direct SLAM (L. von Stumberg, P. Wenzel, Q. Khan and D. Cremers), In preprint, 2019. ([arxiv][video][project page]) [bibtex]
Powered by bibtexbrowser
Export as PDF, TEX or BIB

Rechte Seite

Informatik IX
Chair for Computer Vision & Artificial Intelligence

Boltzmannstrasse 3
85748 Garching

info@vision.in.tum.de