Closed torydebra closed 4 years ago
Probably a fake link can be used to attach a sensor, so we should not obly the user to have a mesh as last link
I change the method used, now it is: https://github.com/ADVRHumanoids/ROSEndEffector/blob/7667964d5d6a1bdf726f08067562cfbc0d75cc4c/include/ROSEndEffector/ParserMoveIt.h#L219-L224
The solution proposed is ok for me.
An issue that comes out for shadow hand, but can be a problem also for others. The shadow in the urdf has as last link for each finger a "virtual" link it has no mesh associated (and also 0 inertia and very low weight). I do not know why there is such a link. So the real tip is the 2nd to last link Anyway, this causes problems to findActions, because for a pinch, a collision between tips can't be found in this case because there is no mesh.
The solution is to specify the finger (in the urdf) putting as last link of the chain (i.e. the tip) the 2nd to last link of the urdf, so the moveitparser identify this as tip. Doing this go against my "definition" of fingertips: https://github.com/ADVRHumanoids/ROSEndEffector/blob/d554ceca31c05ad58cfcfe6d27582645ba44307e/include/ROSEndEffector/ParserMoveIt.h#L219-L222
So I should relax this definition crossing out the last sentence. I am asking if you think this is ok, and also if defining the finger chain in that way can cause problem to parser and rosee online part done by you @liesrock (I think not)