rtabmap

RTAB-Map library and standalone application

3,763

921

3,763

573

View on GitHub

Top Related Projects

ORB_SLAM2

10,153

Real-Time SLAM for Monocular, Stereo and RGB-D Cameras, with Loop Detection and Relocalization Capabilities

maplab

2,836

A Modular and Multi-Modal Mapping Framework

cartographer

7,854

Cartographer is a system that provides real-time simultaneous localization and mapping (SLAM) in 2D and 3D across multiple platforms and sensor configurations.

GTSAM is a library of C++ classes that implement smoothing and mapping (SAM) in robotics and vision, using factor graphs and Bayes networks as the underlying computing paradigm rather than sparse matrices.

Quick Overview

RTAB-Map (Real-Time Appearance-Based Mapping) is an open-source library for SLAM (Simultaneous Localization and Mapping) in robotics and computer vision. It provides real-time 3D mapping and localization capabilities, supporting various sensors and platforms, including RGB-D cameras, stereo cameras, and LiDAR.

Pros

Versatile and supports multiple sensor types and platforms
Real-time performance with efficient loop closure detection
Extensive documentation and active community support
Integration with popular robotics frameworks like ROS

Cons

Steep learning curve for beginners
Resource-intensive for large-scale mapping
Limited support for some specialized sensors
Dependency on external libraries may complicate setup

Code Examples

Initialize RTAB-Map:

#include <rtabmap/core/Rtabmap.h>

rtabmap::Rtabmap rtabmap;
rtabmap::ParametersMap parameters;
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemRehearsalSimilarity(), "0.6"));
rtabmap.init(parameters);

Process RGB-D data:

#include <rtabmap/core/util3d.h>

cv::Mat rgb = cv::imread("rgb.png");
cv::Mat depth = cv::imread("depth.png", cv::IMREAD_UNCHANGED);
rtabmap::SensorData data(rgb, depth, rtabmap::CameraModel());
rtabmap.process(data);

Retrieve and save the map:

std::map<int, rtabmap::Signature> nodes;
std::map<int, rtabmap::Transform> poses;
std::multimap<int, rtabmap::Link> links;
rtabmap.get3DMap(nodes, poses, links);
rtabmap::savePoses("map_poses.txt", poses);

Getting Started

Install dependencies:

sudo apt-get install libpcl-dev libopencv-dev liboctomap-dev

Clone and build RTAB-Map:

git clone https://github.com/introlab/rtabmap.git
cd rtabmap && mkdir build && cd build
cmake ..
make
sudo make install

Include RTAB-Map in your project:

#include <rtabmap/core/Rtabmap.h>
// ... (see code examples above for usage)

Competitor Comparisons

ORB_SLAM2

10,153

Real-Time SLAM for Monocular, Stereo and RGB-D Cameras, with Loop Detection and Relocalization Capabilities

Pros of ORB_SLAM2

Highly efficient and accurate visual SLAM system
Supports monocular, stereo, and RGB-D cameras
Lightweight and suitable for real-time applications on CPU

Cons of ORB_SLAM2

Limited to visual data, no integration with other sensor types
Lacks built-in loop closure and global optimization features
Less active development and community support

Code Comparison

ORB_SLAM2 (C++):

// Feature extraction and matching
void Frame::ExtractORB(int flag, const cv::Mat &im)
{
    (*mpORBextractor)(im,cv::Mat(),mvKeys,mDescriptors);
}

RTAB-Map (C++):

// Feature extraction and matching
void Feature2D::detectAndCompute(
    const cv::Mat & image,
    std::vector<cv::KeyPoint> & keypoints,
    cv::Mat & descriptors)
{
    detector_->detect(image, keypoints);
    extractor_->compute(image, keypoints, descriptors);
}

Both repositories use similar approaches for feature extraction and matching, but RTAB-Map offers more flexibility with separate detector and extractor objects.

maplab

2,836

A Modular and Multi-Modal Mapping Framework

Pros of maplab

Designed for multi-session mapping and localization
Supports visual-inertial odometry and loop closure
Includes tools for map management and optimization

Cons of maplab

Steeper learning curve due to complex architecture
Less active community and development compared to RTAB-Map
Limited support for RGB-D sensors

Code Comparison

maplab:

vi_map::VIMap map;
vi_map::MissionId mission_id;
map.addNewMissionWithBaseframe(mission_id, T_G_M, aslam::Time(0));

RTAB-Map:

rtabmap::Rtabmap rtabmap;
rtabmap::ParametersMap parameters;
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemRehearsalSimilarity(), "0.6"));
rtabmap.init(parameters, "DatabasePath");

Both repositories offer robust SLAM solutions, but maplab focuses on multi-session mapping and visual-inertial odometry, while RTAB-Map provides a more user-friendly approach with broader sensor support. RTAB-Map is generally easier to integrate into existing projects, while maplab offers advanced features for specific use cases in robotics and autonomous systems.

cartographer

7,854

Cartographer is a system that provides real-time simultaneous localization and mapping (SLAM) in 2D and 3D across multiple platforms and sensor configurations.

Pros of Cartographer

Optimized for real-time performance, especially in large-scale environments
Supports both 2D and 3D mapping with various sensor configurations
Provides robust loop closure detection and global optimization

Cons of Cartographer

Steeper learning curve and more complex setup process
Less extensive documentation and community support
Limited integration with other robotics frameworks compared to RTAB-Map

Code Comparison

RTAB-Map (C++):

rtabmap::Rtabmap rtabmap;
rtabmap::ParametersMap parameters;
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemRehearsalSimilarity(), "0.6"));
rtabmap.init(parameters, "MyDatabase.db");

Cartographer (C++):

auto map_builder = cartographer::mapping::CreateMapBuilder(map_builder_options);
auto trajectory_builder = map_builder->AddTrajectoryBuilder(
    trajectory_options, [](const int trajectory_id, const cartographer::common::Time time,
                           const cartographer::transform::Rigid3d& local_pose) {});

Both repositories offer powerful SLAM solutions, with RTAB-Map providing a more user-friendly experience and extensive features, while Cartographer excels in real-time performance and large-scale mapping. The choice between them depends on specific project requirements and user expertise.

Kimera

2,071

Index repo for Kimera code

Pros of Kimera

Focuses on real-time visual-inertial odometry and SLAM, offering high-performance solutions for robotics and AR/VR applications
Implements a modular architecture, allowing easy integration of different components and algorithms
Provides a comprehensive suite of tools for 3D reconstruction, including mesh generation and semantic mapping

Cons of Kimera

Less mature and less widely adopted compared to RTAB-Map
May have a steeper learning curve due to its more specialized focus
Limited documentation and community support compared to RTAB-Map

Code Comparison

RTAB-Map (C++):

rtabmap::Rtabmap rtabmap;
rtabmap::ParametersMap parameters;
parameters.insert(rtabmap::ParametersPair(rtabmap::Parameters::kMemRehearsalSimilarity(), "0.6"));
rtabmap.init(parameters, "MyDatabase.db");

Kimera (C++):

KimeraVIO::Pipeline vio_pipeline(FLAGS_params_folder + "/params");
vio_pipeline.spinOnce();
const gtsam::Pose3& latest_pose = vio_pipeline.getCurrentPose();

Both repositories offer powerful SLAM solutions, with RTAB-Map providing a more general-purpose approach and Kimera focusing on visual-inertial odometry. RTAB-Map has broader adoption and documentation, while Kimera offers specialized features for real-time applications.

gtsam

3,429

Pros of GTSAM

Focuses on factor graph optimization, providing a powerful framework for various robotics and computer vision problems
Offers a wide range of probabilistic inference algorithms and tools
Highly modular and extensible design, allowing for easy integration into existing projects

Cons of GTSAM

Steeper learning curve due to its more abstract and mathematical nature
Less out-of-the-box functionality for complete SLAM solutions compared to RTAB-Map
Requires more manual setup and configuration for specific use cases

Code Comparison

GTSAM example (factor graph creation):

#include <gtsam/nonlinear/NonlinearFactorGraph.h>
#include <gtsam/nonlinear/Values.h>
#include <gtsam/slam/PriorFactor.h>
#include <gtsam/slam/BetweenFactor.h>

gtsam::NonlinearFactorGraph graph;
gtsam::Values initialEstimate;

RTAB-Map example (basic usage):

#include <rtabmap/core/Rtabmap.h>
#include <rtabmap/core/RtabmapThread.h>

rtabmap::Rtabmap rtabmap;
rtabmap::ParametersMap parameters;
rtabmap.init(parameters, "DatabasePath");

Both repositories provide powerful tools for SLAM and mapping, but GTSAM offers a more flexible and general-purpose optimization framework, while RTAB-Map provides a more complete out-of-the-box SLAM solution.