Planning with direct and reverse serial chain
I would like to know if is there a simple way using the MoveIt API to compute the inverse kinematic of the reverse serial chain obtained by switching the base link with the end-effector. Building the reverse URDF tree it is not sufficient because I need both the direct and the reverse serial chains at execution time.
If you are dealing only with usual manipulators it is obviously a nonsense, but imagine to have a serial chain with two grippers attached to the two ends. It would be really useful to call base-link the gripper temporarily secured to the environment and end-effector the free-to-move one (e.g. think of a climbing task). In such a case using MoveIt as usual lets to compute the planning only for one of the two configurations, even if it would be simply a matter of conventions.
At the moment I am exploiting the physics of the environment by planning always with the same end-effector while assigning a proper goal pose to obtain the expected behaviour, but it is clearly not the solution...
