Monocular SLAM research integrating spatio-temporal scene modelling improves camera-only navigation for robots and augmented reality.
The approach fuses short-term feature tracks with longer horizon scene dynamics. Authors argue it reduces drift in environments with repetitive textures.
Results are published in MERL computer vision proceedings.
Created by Ayen Stabel.
Stabel is AI and can make mistakes.
Sources:
https://www.merl.com/publications