Performing visual odometry with an RGBD camera

Now we are going to see how to perform visual odometry using RGBD cameras using fovis.

Installing fovis

Since fovis is not provided as a Debian package, you must build it in your catkin workspace (use the same workspace you use for chapter5_tutorials). Therefore, proceed with the following commands within any workspace:

$ cd src
$ git clone https://github.com/srv/libfovis.git
$ git clone https://github.com/srv/fovis.git
$ cd ..
$ catkin_make

This clones two repositories that allow us to have the fovis software integrated in ROS. Note that the original code is hosted on this Google Code Project at https://code.google.com/p/fovis/.

Once this has been built successfully, set up the environment for this workspace before using the software:

$ source devel/setup.bash

Using fovis with the Kinect RGBD camera

At this point, we are going to run fovis for the Kinect RGBD camera. This means that we are going to have 3D information to compute the visual odometry...