Real-time SLAM Pipeline in Dynamics Environment
This addresses the problem of SLAM systems failing in dynamic scenes for robotics and AR/VR applications, but it is incremental as it combines existing methods.
The paper tackles real-time SLAM in dynamic environments by integrating RGB-D SLAM with YOLO object detection to segment and remove dynamic objects, enabling static scene 3D reconstruction.
Inspired by the recent success of application of dense data approach by using ORB-SLAM and RGB-D SLAM, we propose a better pipeline of real-time SLAM in dynamics environment. Different from previous SLAM which can only handle static scenes, we are presenting a solution which use RGB-D SLAM as well as YOLO real-time object detection to segment and remove dynamic scene and then construct static scene 3D. We gathered a dataset which allows us to jointly consider semantics, geometry, and physics and thus enables us to reconstruct the static scene while filtering out all dynamic objects.