RTAB-Map (Real-Time Appearance-Based Mapping) is a tool for mapping and localization in 3D environments. It uses advanced algorithms to create detailed maps from sensor data such as cameras, lasers, and Kinect. It is compatible with ROS and is used in robotics, augmented reality, and autonomous navigation applications.
- Real-time mapping. RTAB-Map builds 3D maps while the robot or device moves through the environment. It processes sensor data in real time to generate accurate representations of the space, including obstacles and surfaces.
- Appearance-based localization. The system compares visual features of the environment to determine the robot's current position within a previously generated map. This improves accuracy in dynamic environments.
- Multi-sensor integration. Supports RGB-D cameras, LIDAR, depth sensors, and stereo cameras. Data from different sensors is fused to obtain a more complete representation of the environment.
- Loop closure detection. Detects when the robot returns to a location that has already been mapped, correcting accumulated errors in the trajectory. It uses visual descriptors and similarity metrics to identify known places.
- Point cloud processing. Stores and manages large volumes of 3D data, applying filters and optimization algorithms to reduce noise and improve map quality.
- Surface reconstruction. Generates 3D meshes from point clouds, allowing complete structures to be visualized with realistic textures derived from the captured images.
- Map export. Lets you save generated maps in formats such as PLY, OBJ, and PCD, making them easy to use in other 3D visualization or processing software.
- Adjustable memory management. The system manages memory resources through a selective loop closure strategy, prioritizing key areas of the map to maintain stable performance during long sessions.
- ROS support. Integrates with the Robot Operating System (ROS) through dedicated nodes, allowing use on standard robotic platforms and simplifying communication with other system components.
- Integrated visualization. Includes tools for viewing maps, trajectories, and sensor data in real time, with options to customize the graphical display to the needs of your project.
- Autonomous navigation. Provides essential data for path planning systems, including information about obstacles and traversable surfaces, enabling robots to move autonomously.
- Sensor calibration. Offers utilities for calibrating cameras and depth sensors, ensuring that the captured data is accurate and correctly aligned for processing.
RTAB-Map was created by Mathieu Labbé and other contributors at the IntRoLab laboratory at the Université de Sherbrooke in Canada. Development began in 2013 as a solution for real-time mapping on mobile robots. It is written in C++ and uses libraries such as OpenCV, PCL, and Qt. The project remains active, with regular updates that add new features and performance improvements.
