Revision history [back]

The problem of getting the map -> odom transformation is essentially the robot localization problem. So, if you have a map of the environment, you can use a ROS localization package like robot_pose_ekf or AMCL depending on the type of map and the sensor available.

However, if what you need is a simple pose for the robot, you can use the libgazebo_ros_p3d.so gazebo plugin as used here.

It will publish a nav_message/Odometry message to the specified topic. This message essentially is the map -> base_link transformation. You can manipulate this information to publish a /tf yourself. Since ROS doesn't allow multiple parent frames, you can't simply publish a /tf using message_to_tf. You will need to publish map -> odom by subtracting odom -> base_link from map -> base_link. A clear example is given in the AMCL implementation.

The problem of getting the map -> odom transformation is essentially the robot localization problem. So, if you have a map of the environment, you can use a ROS localization package like robot_pose_ekf or AMCL depending on the type of map and the sensor sensors available.

However, if what you need is a simple pose for the robot, you can use the libgazebo_ros_p3d.so gazebo plugin as used here.

It will publish a nav_message/Odometry message to the specified topic. This message essentially is the map -> base_link transformation. You can manipulate this information to publish a /tf yourself. Since ROS doesn't allow multiple parent frames, you can't simply publish a /tf using message_to_tf. You will need to publish map -> odom by subtracting odom -> base_link from map -> base_link. A clear example is given in the AMCL implementation.