Deep LiDAR localization using optical flow sensor-map correspondences.
Anders SunegårdLennart SvenssonTorsten SattlerPublished in: 3DV (2020)
Keyphrases
- optical flow
- inertial navigation
- dense correspondence
- source localization
- map building
- smoothness constraint
- loop closing
- optical flow estimation
- image sequences
- motion estimation
- sensor networks
- real time
- maximum a posteriori
- dense optical flow
- robot localization
- multi sensor
- optical flow computation
- motion model
- high resolution
- motion field
- computer vision
- monte carlo localization
- pairwise
- least squares
- feature tracking
- localization method
- d scene
- multi view
- feature points
- line features
- image features
- camera motion
- correspondence matching
- moving objects
- robot navigation
- multiscale
- rigid body motion
- feature matching