The authors present COIN-LIO, a LiDAR Inertial Odometry (LIO) pipeline that tightly couples information from LiDAR intensity with geometry-based point cloud registration. The focus of their work is to improve the robustness of LiDAR-inertial odometry in geometrically degenerate scenarios, such as tunnels or flat fields.
The key components of their approach are:
Image Processing Pipeline: They project LiDAR intensity returns into an intensity image and propose a filtering method to improve brightness consistency within the image as well as across different scenes.
Geometrically Complementary Feature Selection: They present a novel feature selection scheme that detects uninformative directions in the point cloud registration and explicitly selects patches with complementary image information.
Photometric Error Minimization: They fuse the photometric error minimization in the image patches with inertial measurements and point-to-plane registration in an iterated Extended Kalman Filter.
The authors evaluate their approach on a public dataset (Newer College) and a new dataset they created, called ENWIDE, which captures five real-world environments with long sections of geometrically degenerate scenes. Their results show that the proposed intensity-augmented approach significantly improves accuracy and robustness compared to geometry-only and geometry-and-intensity-based methods, especially in challenging environments where the latter approaches fail.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Patrick Pfre... at arxiv.org 04-26-2024
https://arxiv.org/pdf/2310.01235.pdfDeeper Inquiries