I do have a robotic application, where a 7Dof robot arm is mounted on a omnidirectional mobile platform. My overall goal is to get MoveIt! to calculate a sequence of joint movements, such that the robot EEF reaches a desired goal in Cartesian space.
In order to combine a robot platform with a world, the MoveIt! setup assistant lets you assign virtual joints between the "footprint" of the platform and the world it is placed in.
I do have two strategies. Either
- select a planar joint as a virtual joint. (What are the degrees of freedom or respectively the joint information that I can gather from this joint)
- select a fixed joint and add a (prismatic-x -> prismatic-y -> revolute-z) chain to the robot model.
Are there any significant differences (advantages/ disadvantages) to either of the approaches?