Setting TCP and orientation of endmill

[JMinnesota]

Hello,

I have a Kuka KR60HA with a KRC2, I believe it has system 5.2 or 5.4.  There is a spindle mounted on the robot flange, and we use endmills in the spindle for milling aluminum.  We use this to mill flash off aluminum castings 

The tool center point and orientation were originally set years ago by the robot integrator, I don't know how they set it.

I would like to reset / verify the tool center point and orientation.  I am guessing the orientation is important, because it needs to be perpendicular to the workpiece so the endmills don't cut a draft angle.

In the Kuka manual they have procedures about setting the tool center point and orientation.  To set the tool center point I would probably just put a tapered endmill, or a pointed pin in the spindle, and use the standard procedure for 4 point tcp.

However, how would I set tool orientation?  With a parallel gripper (like the attached picture from the Kuka manual) it is easy to use the edge of the gripper to set the x,y,z tool orientation.  However, with an endmill, the x axis would be in the center of the endmill.  Is there some way to use the edge of the endmill to find tool center point?  Or is there a special pin geometry that I could put in the spindle to set tool orientation?

Also, I am looking for your thoughts on two ways to do this:

1) a high accuraccy, low speed (manual jogging), low cost way to do this

2) a high accuraccy, high speed (automatic) way to do this, that could potentially cost $$$. 

Thanks for the help!

[kykam]

The orientation of the spindle depends on how it's mounted to the flange.  You just need to do a trial and error approach to it.

I'm willing to be that it's mounted perpendicular to the flange or parallel to it.  In that case, you just need to put +-90 on the orientation of one or two fields (A,B,C).  After you do this, move it to a taught point and see if it's in the correct spot.


As for teaching the tool center point, there really isn't any automatic way of doing this.  the 4 point approach is pretty accurate, but putting in CAD data is better.

You could add an arbitrary pointer and teach it as the origin of a base.  At that point maybe put a tube around it.  Create a move that goes to 0,0,0 of the base you taught with the TCP of the endmill.  This will at least verify that your taught TCP is what it was. 

[MeanRobot]

Does your current TCP have an C value (other than 0)?  I would guess not.

So with regard to the orientation, just use the ABC World (5D) method instead of the ABC 2-point method to teach the orientation.

You just have to have the spindle perpendicular to the floor for it to work.

[SkyeFire]

If you have a laser tracker (Faro, Leica, API, etc) and a skilled operator, and some time, you can generate a pretty precise TCP by measuring the physical tool relative to the robot frame. 

The trick is to put the wrist axes all at 0deg, then look for the machined surfaces that you can measure to define the Tool 0 frame (the default TCP in the center of the wrist mounting flange).  Generally, each KUKA wrist has some flats on it between A5 and A6 that can be used to define the direction of the X or Y axis (it varies from model to model).  In addition to that, you need to measure the circle of the mounting flange to define the axis of A6, and then measure the plane of the mounting surface of the flange (often by measuring the back of the tool adapter plate mounted to the flange).  The intersection of the plane and the circle axis defines the position of the Tool 0 frame, and the X/Y flat defines the direction (Z is always + pointing along the A6 axis, out away from the wrist towards the mounted tooling).

Once you've generated that Tool 0 reference, your tracker operator can measure the position and orientation of the tool in any way that s/he likes, and as long as the XYZABC numbers are relative to the Tool 0 flange, the data will be valid (however, some conversion from typical tracker IJK angles into KUKA ABC angles will be required).

[TygerDawg]

see this previous thread
http://www.robot-forum.com/robotforum/general_discussion/calibrating_a_tcp_for_nonphysical_tools-t1224.0.html

if you have CAD model of your EOAT, it is a very simple matter to this mathematically.  You must visualize moving (translating & orienting) the default TCP located at the tool flange & oriented such that the X-Axis is coincident with the dowel hole (...isn't it?  or is it opposite the dowel hole?  whatever...it's defined in the manual), then moving that default TCP step-by-step through the different EOAT components until it is located/oriented where desired.  The in code you build a compound transformation to define the tool.

If I recall correctly, the KUKA syntax would  something like this.

DECL FRAME mytool
mytool = {X nx1, Y ny1, Z nz1, A na1, B nb1, C nc1}:{X nx2, Y ny2, Z nz2, A na2, B nb2, C nc2}:{, , , , , , }: etc
$TOOL = mytool

Alternatively, with CAD, most CAD modellers can give you the coordinate transformation between two coordinate systems.   Put a CS at the flange and working point of your tool model, and get that compound transformation.  Then you'd have to figure out how to change the orienation values to the twisted way KUKA assigns their values.