RTAB-Map

Versatile graph-based SLAM supporting RGB-D, stereo, and LiDAR with real-time operation

License

BSD-3-Clause

Overview

RTAB-Map, short for Real-Time Appearance-Based Mapping, is a graph-based SLAM library that works with RGB-D cameras, stereo cameras and lidar. It provides loop closure detection, memory management for large maps and multi-session mapping, and can export point clouds as PLY or PCD. It integrates with ROS 1 and ROS 2 and is used for mobile robot navigation and 3D reconstruction.

RTAB-Map (Real-Time Appearance-Based Mapping) is a versatile graph-based SLAM library that supports various sensor configurations. It's designed for real-time operation and large-scale mapping.

Key features include: - RGB-D, stereo, and lidar support - Real-time operation - Memory management for large maps - Loop closure detection - Multi-session mapping - ROS 1/2 integration - Visualization tools - Cloud export (PLY, PCD)

RTAB-Map is popular for mobile robot navigation and 3D reconstruction. Its ROS integration makes it easy to use with standard robot platforms.

Why it matters

RTAB-Map is the most versatile SLAM solution in the ROS ecosystem. Supporting RGB-D cameras, stereo cameras, and LiDAR, it provides reliable real-time mapping and localization that works across different sensor configurations — critical for mobile robot deployment.

Problem it solves

Robots need to build maps of their environment and localize within them, but different applications require different sensors. RTAB-Map provides a single SLAM solution that works with whatever sensors are available.

Key takeaways

• Multi-sensor SLAM supporting RGB-D, stereo, and LiDAR
• Real-time loop closure detection for drift correction
• Large-scale mapping with memory management for long-duration operation
• Native ROS and ROS 2 integration with visualization tools
• Actively maintained with regular updates and improvements

Who it is for

• Mobile robot developers needing SLAM capabilities
• Teams building mapping and exploration robots
• Researchers comparing SLAM algorithms
• Companies deploying robots in indoor environments

Limitations

• Computationally intensive — requires decent CPU/GPU for real-time operation
• Parameter tuning needed for optimal performance in different environments
• Dense mapping generates large datasets requiring storage management
• Outdoor SLAM less reliable than indoor due to feature matching challenges