koide3/small_gicp

Health v2 · maintenance only

Adoption not part of the score

1023 stars · 129 forks observed · 2026-09-01

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

small_gicp is a header-only C++ library with Python bindings for fast, parallelized point cloud registration algorithms including ICP, Point-to-Plane ICP, GICP, and VGICP. It is an optimized rewrite of fast_gicp with minimal dependencies (Eigen, bundled nanoflann and Sophus) and OpenMP/Intel TBB parallelism backends.

Use cases

• align two 3d point clouds with icp in c++
• fast gicp scan matching for lidar slam
• python point cloud registration with open3d
• drop-in replacement for pcl iterative closest point
• parallel kd-tree construction and point cloud downsampling
• register 3d scans for robotics odometry
• multithreaded point cloud alignment library

When to choose

• you need CPU-based point cloud registration with high performance and parallelism
• you want minimal dependencies and a header-only C++ integration
• you need Python bindings that work with Open3D without PCL
• you want a customizable registration pipeline via traits and templates

When to avoid

• you need GPU-accelerated registration - this package has no GPU implementations
• you require coarse/global registration rather than fine local alignment
• your project is not C++ or Python

Facets

library · maturity active

computer-vision image-processing math concurrency robotics computer-vision autonomous-vehicles cpp python windows cross-platform point-cloud-registration icp gicp scan-matching lidar slam header-only openmp tbb pcl open3d kd-tree algorithms linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="koide3/small_gicp")