Expand description
Iterative Closest Point (ICP) SLAM implementation
This module implements the ICP algorithm for point cloud registration. It finds the optimal transformation (rotation and translation) between two point sets by iteratively minimizing the distance between corresponding points.
Ported from PythonRobotics Original authors: Atsushi Sakai (@Atsushi_twi), Göktuğ Karakaşlı, Shamil Gemuev
Structs§
- ICPResult
- ICPTrace
Frame - One accepted ICP alignment, including the initial alignment. The points, nearest-neighbor indices and error all describe this same state, before the next rigid fit. Rejected divergent updates are not included.
Functions§
- apply_
2d_ transformation - Apply 2D transformation to points
- generate_
2d_ points - Generate random 2D point cloud for testing
- icp_
matching - Main ICP matching function
- icp_
matching_ with_ trace - Run the same solver as
icp_matching, recording its accepted states. Inputs have the same requirements: nonempty finite 2xN point matrices. Frame zero is the input cloud with identity transform; the last frame describes the returned result. The legacyiterationscount also includes an attempted assessment that rejects a divergent fit.