This post provides a brief overview of how to get a basic visual odometry demo up and running using a Multisense-SL. This demo requires the installation of the viso2 stack and was tested on Ubuntu 12.04 64 bit install using ROS Fuerte. (Note: The demo will also work on Groovy provided the groovy viso2 branch is built and installed). A full listing of the commands to run this demo can be found at the end of this post.
The viso2 ROS stack is a wrapper of the libvios2 library developed by Andreas Geiger. It runs by matching features between left and right rectified images at time t and t-1.
To install the viso2 stack checkout the fuerte branch into the same workspace where the Multisense-SL driver is installed. After building the viso2 stack copy the attached multisense_viso.launch file into the viso2/viso2_ros/launch directory.
Next launch the Multisense-SL driver and then the viso2 visual odometry driver. After both drivers have been successfully launched rosrun RViz and load the attached viso2_rviz_config.vcg RViz configuration file.
To visualize the laser_scan point clouds in RViz increase the decay time of the laser_scan topic.
(Note images were taking with the laser stationary.)
rosws init . /opt/ros/fuerte
rosws set multisense --git https://github.com/carnegierobotics/multisense_ros
rosws update multisense
git clone https://github.com/srv/viso2.git -b fuerte
rosws set viso2
After downloading the attached multisense_viso.launch and viso2_rviz_config in the $HOME/Downloads directory
cp ~/Downloads/multisense_viso.launch ~/viso_demo/viso2/viso2_ros/launch/
cp ~/Downloads/viso2_rviz_config.vcg ~/viso_demo/viso2/
roslaunch multisense_bringup multisense.launch
In a new terminal
roslaunch viso2_ros multisense_viso.launch
In a new terminal
rosrun rviz rviz -d ~/viso2/viso2_rviz_config.vcg