Skip to content

VINS-Fusion based visual-inertial SLAM with tightly-coupled wheel odometry for outdoor delivery robot

License

Notifications You must be signed in to change notification settings

KopanevPavel/runbot_custom_localization

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

29 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VINS-Fusion based SLAM with tightly-coupled wheel odometry and DNN based front-end

Localization package for outdoor delivery robot [Skoltech ISR Lab project]

DNN based front-end

Deep neural network-based front-end and offline optimization thread were implemented in the package. The architecture of the proposed system is shown below:

SLAM front-end is based on two state-of-the-art feature extraction and matching DNN architectures: SuperPoint and SuperGlue. SuperPoint and SuperGlue neural networks on NVIDIA Jetson AGX Xavier had 1.13 seconds mean forward pass time. Therefore DNN based front-end was utilized in an offline manner. The agent's position was estimated in the parallel thread.

Graph based back-end

Tightly-coupled wheel odometry preintegration approach is proposed, which adds additional factors into the optimization pose estimation problem and improves pose estimation accuracy. System architecture:

About

VINS-Fusion based visual-inertial SLAM with tightly-coupled wheel odometry for outdoor delivery robot

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages