Dual Grinder/Sander E..O.A.T. concerns?
|Not answered yet|
We are programming our UR10 to perform repetitive deburr grinding and sanding operations. We are able to accomplish each operation separately; using Force Control on a Robotiq 2F Gripper that holds the grinder/sander separately.
We are looking into either having the robot switch out its tooling without human intervention, or have it hold both tools at the same time. Both options seem to present obstacles.
We can create a jig where the robot is able to place the tools down and pick the tools up from the same location using a 'tool change' script/program. Then the obstacles would be producing a repeatable result where it is able to grab the tool securely. [The gripper is too weak to hold the tool in place, so currently the gripper has extremely tight guides that the fingers fit in between. It is very difficult\nearly impossible to have the robot grab the current tooling without human intervention, wiggling the part into place.]
If we had the robot hold both tools at the same time, my concern would be force control's ability to maintain accuracy.
Currently, when the grinder or sander is operating, the centrifugal force of the tool, when activated, will disrupt the force control's 'adaptive stiffness' settings. As long as adaptive stiffness is active, the results of force control are random. Sometimes it will make contact with the surface, and sometimes it moves in the opposing direction, with no contact or force being applied to the E.O.A.T. or arm to justify its retreat. (I suspect it has to do with a shift in weight or TCP, not being zero'd appropriately after tool is activated?) This can be 'fixed' by manually inputting a stiffness setting of my choice; according to the amount of force being applied and the articulation of the movements I need to make.
My concern with having two tools on one arm\force-sensor is that if I add another tool to the same arm, it will complicate the force control even more. With two tools, rather than applying force in a z direction as we have been doing, we would either be applying force in an X or Y direction, and most likely some combination of x, y, and z.
Does anybody have any experience with similar applications?