Skip to main content

​
What is ORB-SLAM3?

ORB-SLAM3 is the first real-time SLAM library able to perform Visual, Visual-Inertial and Multi-Map SLAM with monocular, stereo and RGB-D cameras, using pin-hole and fisheye lens models. In all sensor configurations, ORB-SLAM3 is as robust as the best systems available in the literature, and significantly more accurate.
ORB-SLAM3 v1.0 was released in December 2021 by the SLAM Lab at the University of Zaragoza.

​
Key Features

​
Multiple Sensor Configurations

ORB-SLAM3 supports a wide range of sensor configurations:
  • Monocular: Single camera SLAM
  • Stereo: Dual camera stereo vision
  • RGB-D: Depth camera (e.g., Kinect, RealSense D435i)
  • Monocular-Inertial: Single camera with IMU
  • Stereo-Inertial: Dual camera with IMU
  • RGB-D-Inertial: Depth camera with IMU

​
Camera Model Support

  • Pin-hole cameras: Standard perspective projection cameras
  • Fisheye cameras: Wide-angle fisheye lens models

​
Advanced Capabilities

  • Real-time performance: Processes frames at camera frame rate on modern hardware
  • Multi-map SLAM: Ability to build and manage multiple maps simultaneously
  • Loop closure: Automatic detection and correction of accumulated drift
  • Relocalization: Recovery after tracking loss
  • Map reuse: Save and load maps for later use
  • Visual-inertial initialization: Robust initialization using IMU data

​
Use Cases and Applications

ORB-SLAM3 is designed for a wide range of robotics and computer vision applications:
  • Autonomous Navigation: Robot localization and mapping in unknown environments
  • Augmented Reality: Real-time camera pose estimation for AR applications
  • Drones and UAVs: Visual-inertial odometry for aerial vehicles
  • Mobile Robotics: Indoor and outdoor navigation for wheeled robots
  • Research: Benchmark and baseline for SLAM algorithm development

​
Architecture Overview

ORB-SLAM3 consists of three main parallel threads:
  1. Tracking: Localizes the camera with every frame and decides when to insert new keyframes
  2. Local Mapping: Processes new keyframes and performs local bundle adjustment
  3. Loop Closing: Detects large loops and performs pose-graph optimization
The system uses:
  • ORB features: Fast and robust binary features for matching
  • DBoW2: Place recognition for loop detection and relocalization
  • g2o: Graph optimization framework for bundle adjustment

​
Performance

ORB-SLAM3 has been tested on popular datasets:
  • EuRoC dataset: MAV with stereo cameras and IMU (pin-hole lenses)
  • TUM-VI dataset: Handheld device with stereo fisheye cameras and IMU
  • KITTI dataset: Automotive stereo camera setup
  • TUM RGB-D dataset: RGB-D camera sequences
A powerful computer (e.g., Intel i7 or equivalent) is recommended to ensure real-time performance and more stable, accurate results.
ORB-SLAM3 is built on years of research in Visual SLAM:
  • [ORB-SLAM3] Carlos Campos, Richard Elvira, Juan J. Gómez Rodríguez, José M. M. Montiel and Juan D. Tardós, ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual-Inertial and Multi-Map SLAM, IEEE Transactions on Robotics 37(6):1874-1890, Dec. 2021. PDF
  • [IMU-Initialization] Carlos Campos, J. M. M. Montiel and Juan D. Tardós, Inertial-Only Optimization for Visual-Inertial Initialization, ICRA 2020. PDF
  • [ORBSLAM-Atlas] Richard Elvira, J. M. M. Montiel and Juan D. Tardós, ORBSLAM-Atlas: a robust and accurate multi-map system, IROS 2019. PDF
  • [ORBSLAM-VI] Raúl Mur-Artal, and Juan D. Tardós, Visual-inertial monocular SLAM with map reuse, IEEE Robotics and Automation Letters, vol. 2 no. 2, pp. 796-803, 2017. PDF
  • [ORB-SLAM2] Raúl Mur-Artal and Juan D. Tardós. ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras. IEEE Transactions on Robotics, vol. 33, no. 5, pp. 1255-1262, 2017. PDF

​
License

ORB-SLAM3 is released under the GPLv3 license. For commercial use, please contact the authors at orbslam (at) unizar (dot) es.

​
Citation

If you use ORB-SLAM3 in an academic work, please cite: 
@article{ORBSLAM3_TRO,
  title={{ORB-SLAM3}: An Accurate Open-Source Library for Visual, Visual-Inertial 
           and Multi-Map {SLAM}},
  author={Campos, Carlos AND Elvira, Richard AND G\'omez, Juan J. AND Montiel, 
          Jos\'e M. M. AND Tard\'os, Juan D.},
  journal={IEEE Transactions on Robotics}, 
  volume={37},
  number={6},
  pages={1874-1890},
  year={2021}
}

​
Next Steps

Install ORB-SLAM3

Set up the library and its dependencies on your system

Quick Start

Run your first SLAM example in minutes

Build docs developers (and LLMs) love

Get started for freeTalk to us