Learning-AI

Multi-object Monocular SLAM for Dynamic Environments

June 2020

tl;dr: Multi object SLAM with keypoint.

Overall impression

“Multibody mono SLAM” comes from “multibody SfM”, and means it not only tracks camera motion but also the motion of the dynamic rigid objects in the scene.

Conventional vSLAM discards dynamic objects in the scene, multibody vSLAM explicitly estimates the pose of objects.

Key ideas

Technical details

Notes