Example usage of the tsdf_localization package.

Requirements:

• ROS-noetic
• tsdf_localization
• gazebo
• robot_pose_ekf

Clone this repository into your ROS(-noetic) workspace:

$ git clone https://github.com/uos/tsdf_localization_demo.git

Build your workspace.

Go to tsdf_localization_demo/maps-folder and enter

$ bunzip2 avz_tsdf.h5.bz2

Start a simulated robot by calling:

$ roslaunch tsdf_localization_demo start_robot.launch

A window should open that looks like this:

TODO: image of gazebo

Next, run the TSDF MCL by calling:

$ roslaunch tsdf_localization_demo mcl_3d.launch

The launch file starts

• mcl_3d
• visualization:
  • RViz
  • tsdf_vis: h5 -> marker

It initially starts with an equally distributed belief state. Steer the robot until the particles are converging. Similar to amcl, you can initialize TSDF-MCL during runtime either by guessing the initial pose using RViz:

or by calling the service /global_localization:

rosservice call /global_localization