Skip to content

Latest commit

 

History

History
15 lines (12 loc) · 690 Bytes

kn-slam.md

File metadata and controls

15 lines (12 loc) · 690 Bytes

KN-SLAM: Keypoints and Neural Implicit Encoding SLAM

Author: Wu

Year: 2024

Notes:

  • SLAM with implicit representation show poor localization and no loop closure with RGBD
  • Fused feature based localization with loop closure and implicit representation (the one of NICESLAM)
  • IMAP uses a single mlp to represent the entire scene
  • Nice slam uses multiple encoder-decoder pair at different resolution for scene representation
  • vote KF with overlap ratio
  • perform joint optimization on reprojection, depth and color loss to refine pose and scene representation parameters
  • loop closure with simple kp matching and pose graph optimization + refine scene representation parameters