Path planning for mobile navigation is typically done by move_base
as part of the navigation stack. MoveIt on the other hand is usually used for n-DOF manipulator path planning (including 3D-occupancy maps and collision-avoidance).
So for mobile robots you have a move_base
node differentially driving the base to a place (often using SLAM with laser scanners). Once the base goal is done, a trajectory for the 6-/7-axis (multi-)arm is planned in a high dimensional space: 3D-cameras (e.g. stereo pairs, structured light or time-of-flight sensors) perceive the environment and OMPL or similar planners come into place to generate trajectories for the manipulators while the base stands still.
What do you want to achieve for 2D-setups using OMPL?
Similar question without answer: https://answers.ros.org/question/2780...