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

by   Lukas von Stumberg, et al.

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.


page 4

page 6

page 7

page 12

page 13


Training Deep SLAM on Single Frames

Learning-based visual odometry and SLAM methods demonstrate a steady imp...

Online Photometric Calibration for Auto Exposure Video for Realtime Visual Odometry and SLAM

Recent direct visual odometry and SLAM algorithms have demonstrated impr...

FDMO: Feature Assisted Direct Monocular Odometry

Visual Odometry (VO) can be categorized as being either direct or featur...

Evaluation of the visual odometry methods for semi-dense real-time

Recent decades have witnessed a significant increase in the use of visua...

ObjectMatch: Robust Registration using Canonical Object Correspondences

We present ObjectMatch, a semantic and object-centric camera pose estima...

A Unified Formulation for Visual Odometry

Monocular Odometry systems can be broadly categorized as being either Di...

DynaPix SLAM: A Pixel-Based Dynamic SLAM Approach

In static environments, visual simultaneous localization and mapping (V-...

Please sign up or login with your details

Forgot password? Click here to reset