Only released in EOL distros:  

rgbdslam_freiburg: rgbdslam

Package Summary

This package can be used to register the point clouds from RGBD sensors such as the kinect or stereo cameras. The rgbdslam node can be connected easily to an octomap_server node to create a memory-efficient 3D map.

Package Summary

rgbdslam (v2) is a SLAM solution for RGB-D cameras. It provides the current pose of the camera and allows to create a registered point cloud or an octomap. It features a GUI interface for easy usage, but can also be controlled by ROS service calls, e.g., when running on a robot. For installation and usage instructions see the README file of the github repository.

  • Maintainer status: maintained
  • Maintainer: Felix Endres <endres AT informatik.uni-freiburg DOT de>
  • Author: Felix Endres, Juergen Hess, Nikolas Engelhard
  • License: GPLv3
  • Source: git (branch: indigo)


This page is about the RGB-D SLAM Version for ROS Electric, i.e. the subversion repository revisions up to #2924 (checkout via svn co URL@REVISION. It succeeds our entry at the ROS 3D contest (see the contest page).

See rgbdslam/evaluation for instructions on how to install and evaluate RGB-D SLAM with the RGB-D Benchmark Dataset


RGBDSLAM has been developed and tested on Ubuntu 11.10 featuring ROS electric, but should work on earlier Ubuntus and ROS diamondback with minor modifications. If you encounter problems, please ask at Feel free to send bug fixes, usage reports and suggestions to endres(at)


The detailed preconditions are:

  • Ubuntu 11.10, 64bit Version (see your /etc/issue and the output of uname -m)

  • ROS electric is installed
  • Qt4 installed as default Qt version (check/change with "update-alternatives --display qmake")
  • Optional: OpenGL Hardware Support, required for visualization and SIFTGPU

Installation should work with (but hasn't been tested)

  • Ubuntu 11.04 and 10.10
  • ROS diamondback (except ORB features)

If you do not already have a directory for personal ROS packages, create one like this:

mkdir -p ~/ros
echo 'export ROS_PACKAGE_PATH=~/ros:$ROS_PACKAGE_PATH' >> ~/.bashrc
source ~/.bashrc

The remainder of this description assumes you chose ~/ros/ as your personal package path. Otherwise you need to adapt the paths below, where ~/ros is contained to fit your directory strucure.

RGBDSLAM (+ Requirements)

Download RGBDSLAM and g2o to your personal package path:

svn co
svn co ~/ros/g2o

Build rgbdslam and g2o.

rosmake --rosdep-install rgbdslam

This takes quite a time (20min on an old laptop with core2 cpu T5600 at 1.83GHz). You can advance to the next section and fetch the benchmark data and tools meanwhile. Using "export ROS_PARALLEL_JOBS=-jX" with X being the number of your cpu cors at maximum, will speed the compilation up by a factor of X, but will need huge amounts of memory (~ 1GB per core).

If ros dependency resolution failed for your system for some reason, you might need to install the system dependencies manually:

sudo apt-get install libglew1.5-dev libdevil-dev libsuitesparse-dev

If this fails and you can not figure out the solution, please check/ask for solutions at

Optionally, you might want to install perception_pcl_addons (debian package: ros-diamondback-perception-pcl-addons) which includes pcd_viewer, a viewer for the file format used by RGBDSLAM to store registered point clouds. If you want to process the point cloud you can save to a ply file and use meshlab (debian-package: meshlab).


To start RGBDSLAM launch

roslaunch rgbdslam kinect+rgbdslam.launch

Alternatively you can start the openni nodes and RGBDSLAM separately, e.g.:

roslaunch openni_camera openni_node.launch
rosrun rgbdslam rgbdslam

To capture models either press space to start recording a continuous stream or press enter to record a single frame. To reduce data redundancy, sequential frames from (almost) the same position are not included in the final model. The 3D visualization always shows the globally optimized model. Neighbouring points are triangulated except at missing values and depth jumps.


To generate octomaps from the registered point clouds you need to install the octomap stacks. The easiest way (for non-colored octomaps) ist to install from the debian repositories:

sudo apt-get install ros-electric-octomap-mapping ros-electric-octomap-visualization

Generating Colored Octomap requires the experimental version of the octomap stack to be installed.

  • Checkout and compile the experimental version of the octomap stack:

svn co
  • Start rgbdslam as described above, capture a scene
  • Execute roslaunch rgbdslam octomap_server.launch

  • In rgbdslam select *Graph->Send Model* (or press Ctrl+M)

  • Execute rosrun octomap_server octomap_saver <>

The parameters in octomap_mapping.launch allow for some fine-tuning of resolution and sensor range. You can use rosrun octovis octovis <> to view the octomap.

Common Issues

Here are some known issues:

  1. If you have issues compiling, make sure you removed any older version of RGBDSLAM from your $ROS_PACKAGE_PATH
  2. RGBDSLAM needs electric or diamondback. Make sure that neither your $PATH nor $ROS_PACKAGE_PATH contains "boxturtle", "cturtle" or "unstable".
  3. If you want to use rgbdslam with fuerte have a look at, where some people describe how they successfully made it run on fuerte.

  4. If you are located in Germany and get errors loading the saved ply files into meshlab, try switching to U.S. locale or replace the decimal point with a comma in your .ply file
  5. To speed up compile times consider to use "export ROS_PARALLEL_JOBS=-j<#cpus>" before rosmake, but you should have lots of memory as gcc may take up to 2GB for four parallel jobs.

  6. If the rosmake process fails with error "fatal error: gsl/gsl_linalg.h: No such file" you will need to install libgsl "sudo apt-get install gsl-bin libgsl0-dev"

Further Information

...can be found in the README file in the RGBDSLAM package.


Nikolas Engelhard and Felix Endres and Jürgen Hess and Jürgen Sturm and Wolfram Burgard, "Real-time 3D visual SLAM with a hand-held RGB-D camera", Proc. of the RGB-D Workshop on 3D Perception in Robotics at the European Robotics Forum, Vasteras, Sweden, 2011. pdf

@InProceedings{ engelhard11euron-workshop,
    TITLE = "Real-time 3D visual SLAM with a hand-held RGB-D camera",
    AUTHOR = "Nikolas Engelhard and Felix Endres and J{\"u}rgen Hess and J{\"u}rgen Sturm and Wolfram Burgard",
    BOOKTITLE = "Proc. of the RGB-D Workshop on 3D Perception in Robotics at the European Robotics Forum",
    ADDRESS = "Vasteras, Sweden",
    YEAR = "2011",
    MONTH = "April",

Felix Endres and Jürgen Hess and Nikolas Engelhard and Jürgen Sturm and Daniel Cremers and Wolfram Burgard, "An Evaluation of the RGB-D SLAM System", In the Proc. of the International Conference on Robotics and Automation (ICRA) 2012. pdf

author={F. Endres and J. Hess and N. Engelhard and J. Sturm and D. Cremers and W. Burgard},
title={An Evaluation of the {RGB-D SLAM} System},
        booktitle = "Proc. of the IEEE Int. Conf. on Robotics and Automation (ICRA)",
address={St. Paul, MA, USA},

Wiki: rgbdslam_electric (last edited 2012-10-31 10:10:19 by Felix Endres)