# koide3/glim

GLIM: versatile and extensible point cloud-based 3D localization and mapping framework

Repository: https://github.com/koide3/glim
Canonical: https://ross.abutalabs.com/products/glim
Homepage: https://koide3.github.io/glim/
Language: C++
License: MIT
License Family: permissive
Topics: slam, 3d, gpu, lidar, mapping, ros, imu, localization, ros2
Last push: 2026-08-17T22:49:18+00:00

## Health v2 (maintenance only)
Score: 76/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 98, release rhythm 35, longevity 100
- inputs: {"age_days": 1748, "days_push": 16, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1758, forks 272 (observed 2026-08-28T04:05:32.393387+00:00)

## What it is
GLIM is a versatile and extensible point cloud-based 3D localization and mapping (SLAM) framework written in C++. It performs direct multi-scan registration on factor graphs with optional GPU acceleration and supports a wide range of range sensors including spinning, non-repetitive, and solid-state LiDARs and RGB-D cameras.

## Use cases
- build 3D maps from spinning LiDAR data like Velodyne or Ouster
- run LiDAR-inertial odometry and mapping with Livox MID360
- map indoor environments with an RGB-D camera or RealSense L515
- correct mapping failures interactively and refine 3D maps
- extend SLAM with custom loop closure or LiDAR-visual-inertial constraints
- run real-time 3D localization on NVIDIA Jetson for a robot
- merge multiple mapping sessions into a single consistent map

## When to choose
- you need accurate factor-graph-based 3D LiDAR or RGB-D SLAM with GPU acceleration
- you want a sensor-agnostic mapping framework that works with many range sensor types
- you need interactive map correction and extensibility via callback slots and extension modules
- you are deploying SLAM on Ubuntu with CUDA or NVIDIA Jetson hardware

## When to avoid
- you need visual-only or 2D SLAM rather than range-based 3D mapping
- you require a pure CPU pipeline without CUDA dependencies and want GPU-free simplicity
- you need a turnkey commercial SLAM product rather than a research-oriented C++ framework
- your platform is Windows or macOS without Linux/CUDA support

## Facets
- artifact type: framework
- maturity: active
- function: computer-vision, graphics, simulation, gpu-computing, middleware
- domain: robotics, autonomous-vehicles, computer-vision, developer-tools
- platform: cpp
- tags: slam, lidar, point-cloud, mapping, localization, factor-graph, imu, ros2, 3d-mapping, odometry, linux, ros, gpu, docker

## Member repositories
- koide3/glim (main) score 76

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:32.393387+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:27:34.785995+00:00, confidence not recorded.
  - readme: https://github.com/koide3/glim (fetched 2026-08-28T04:05:32.393387+00:00, sha bb41941b1fff)
  - homepage: https://koide3.github.io/glim/ (fetched 2026-08-29T11:05:41.512146+00:00, sha 28136c68fc4f)
- Data as of 2026-08-30T08:39:29.467469+00:00.