For the RTAB-Map libraries and standalone application, visit [RTAB-Map's home page](http://introlab.github.io/rtabmap) or [RTAB-Map's wiki](https://github.com/introlab/rtabmap/wiki).
* For armhf architecture, `ros-indigo-rtabmap-ros` is not available. Install `ros-indigo-rtabmap` and build from source `rtabmap_ros` using the `indigo-devel` branch.
* Note that rtabmap_ros Hydro binaries are stuck at version 0.8.12. To use the latest version, see [Build from source](https://github.com/introlab/rtabmap_ros#build-from-source) below.
When launching `rtabmap_ros`'s nodes, if you have the error `error while loading shared libraries...`, add the next line at the end of your `~/.bashrc` to fix it:
This section shows how to install RTAB-Map ros-pkg on **ROS Hydro/Indigo/Jade/Kinetic/Lunar/Melodic** (Catkin build). RTAB-Map works only with the PCL >=1.7, which is the default version installed with ROS Hydro/Indigo/Jade/Kinetic/Lunar/Melodic (**Fuerte and Groovy are not supported**).
* The next instructions assume that you have set up your ROS workspace using this [tutorial](http://wiki.ros.org/catkin/Tutorials/create_a_workspace). I will use melodic prefix for convenience, but it should work with Hydro, Indigo, Jade, Kinetic and Lunar. The workspace path is `~/catkin_ws` and your `~/.bashrc` contains:
* If you want SURF/SIFT on Indigo/Jade (Hydro has already SIFT/SURF), you have to build [OpenCV]([OpenCV](http://opencv.org/)) from source to have access to *nonfree* module. Install it in `/usr/local` (default) and the rtabmap library should link with it instead of the one installed in ROS.
* On Indigo/Jade, I recommend to use latest 2.4 version ([2.4.11](https://github.com/Itseez/opencv/archive/2.4.11.zip)) and build it from source following these [instructions](http://docs.opencv.org/doc/tutorials/introduction/linux_install/linux_install.html#building-opencv-from-source-using-cmake-using-the-command-line). RTAB-Map can build with OpenCV3+[xfeatures2d](https://github.com/Itseez/opencv_contrib/tree/master/modules/xfeatures2d) module, but rtabmap_ros package will have libraries conflict as cv-bridge is depending on OpenCV2. If you want OpenCV3, you should build ros [vision-opencv](https://github.com/ros-perception/vision_opencv) package yourself (and all ros packages depending on it) so it can link on OpenCV3.
* On Kinetic/Lunar/Melodic, I recommend to use OpenCV3+[xfeatures2d](https://github.com/Itseez/opencv_contrib/tree/master/modules/xfeatures2d) module already installed by ROS. You can also install OpenCV2, but rtabmap_ros package will have libraries conflict as cv-bridge is depending on OpenCV3. Thus if you want OpenCV2 on Kinetic/Lunar/Melodic, you should build ros [vision-opencv](https://github.com/ros-perception/vision_opencv) package yourself (and all ros packages depending on it) so it can link on OpenCV2.
* g2o: Use directly the binaries `ros-melodic-libg2o`. However, [this g2o version](https://github.com/felixendres/g2o/tree/c++03) (c++03 branch) built from source may be faster than the binaries (install `libsuitesparse-dev` before building `g2o`) and would be [required to avoid some crashes](http://official-rtab-map-forum.67519.x6.nabble.com/ROS-2D-occupancy-grid-tp1204p1215.html). To build RTAB-Map against [latest official g2o version](https://github.com/RainerKuemmerle/g2o) built from source, g2o should be built with `-DBUILD_WITH_MARCH_NATIVE=OFF` to avoid some segmentation faults caused by Eigen.
* [GTSAM](https://collab.cc.gatech.edu/borg/gtsam): Follow installation instructions from [here](https://collab.cc.gatech.edu/borg/gtsam/#quickstart). RTAB-Map needs latest version from source, it will **not build** with 3.2.1. Set cmake variable `GTSAM_USE_SYSTEM_EIGEN=ON` to make sure the same Eigen version is used across all dependencies to avoid crashes.
2. Install RTAB-Map standalone libraries. Add `-DCMAKE_INSTALL_PREFIX=~/catkin_ws/devel` to `cmake` command below if you want to install in your Catkin's devel folder without `sudo`. **Do not clone in your Catkin workspace**.
These instructions are for Jetpack 3 (Ubuntu 16.04 with ROS Kinetic). For **Jetpack 4** (Ubuntu 18.04 with ROS Melodic), see this [post](https://github.com/introlab/rtabmap/issues/427#issuecomment-608052821).
To use `rtabmap_ros` on Jetson, you can follow the instructions above if you don't care if OpenCV is built for Tegra. However, if you want `rtabmap` to use OpenCV 4 Tegra, we must re-build [vision_opencv](https://github.com/ros-perception/vision_opencv) stack from source too to avoid conflicts with [vision_opencv](https://github.com/ros-perception/vision_opencv) stack binaries from ros (which are linked on a not optimized version of OpenCV). Here are the steps:
1. Install [JetPack](https://developer.nvidia.com/embedded/jetpack) with OpenCV on the Jetson.
2. Do steps 1.2 and 1.3 from http://wiki.ros.org/kinetic/Installation/Ubuntu
9. Install RTAB-Map standalone libraries. Add `-DCMAKE_INSTALL_PREFIX=~/catkin_ws/devel` to `cmake` command below if you want to install in your Catkin's devel folder without `sudo`. **Do not clone in your Catkin workspace**.
10. Clone [vision_opencv](https://github.com/ros-perception/vision_opencv), [image_transport_plugins](https://github.com/ros-perception/image_transport_plugins) and `rtabmap_ros` packages in your catkin_ws: