points teached VS calculated

s084706

Hi all

I have a question regarding deviations that occur when moving to the pick point.
I checked the load of the EOAT and the TCP and these all look good, but still I have some deviations when picking the item.
It’s maybe a stupid thing, but I need to make sure.
I know that robots can’t handle calculated points quite as good as teached points, but I wonder:
when I teach a point and do a offset on this, isn’t this the same as a normal teached point?

For example:
pPick:=Offs(pPick,0,0,100)
MoveJ pPick,vSlow,z5,tGripper;
is this now a less accurate point? Because for the MoveJ instructions the pPick point has fixed values or is this not true?
I can imagine that following instruction is indeed harder for the robot
MoveJ Offs(pPick,0,0,100),vSlow,z5,tGripper;
but I want to know if the first one is also inaccurate. or do we always need to make the instruction as below if we want to make sure that the robot handles it well.
MoveJ pPick,vSlow,z5,tGripper;

thanks for all the help!

Micky

Hello,
I think you should use the argument "fine" instead of "z5" in your Move-Instruction if you want to use an accurate position.

lemster68

I agree with Micky and why not MoveL?

s084706

Hi Micky and Lemster68,

I completely agree with what you say! To give you the truth I was writing on my phone and in the quickness of my action I wasn't thinking about these things.
My code at the moment is this:

        rPickPos_CalculatePickPos;

        ...

        TriggJ pPick_Approach,vHighSpeed,triggPickStart,z10,tGripper\WObj:=wobj0;

        ConfL\Off;

        SingArea\Wrist;

        TriggL pPick,vLowSpeed,triggVacuumON,fine\inpos:=spdPickStopData,tGripper\WObj:=wobj0;
        ...

In the rPickPos_CalculatePickPos routine I have this:

        pPick:=Comm.Result.rtResultTarget;

        pPick_Approach:=RelTool(pPick,0,0,-40);

        pPick_Retract:=RelTool(pPick,0,0,-40);

And it's at the TriggL-command where I have deviations.

lemster68

So you are near singularity?  That plus SingAreaWrist would probably account for inaccuracy.  If you have newer robotware you could try some different optional argument like lock axis 4.  Check on those in the manual.

s084706

Yes, sometimes I'm near singularity. Not always, the pick pos changes.
I know the optional argument you are talking about, but this is not preferable because Another orientation isn't possible.
do you know if there is a difference between the calculated and teached points (as mentioned in my first post)?

lemster68

Not that I am aware of.  ConL/Off could also be part of the problem if it is sometimes complete opposite configurations, I know that from experience.

lemster68

Can you teach your pick with axis 4 at 0 or 180 degrees?  Also, have you tried it without the optional argument Wrist?

s084706

lemster68 said:

Not that I am aware of.  ConL/Off could also be part of the problem if it is sometimes complete opposite configurations, I know that from experience.

What do you mean exactly by that?

lemster68 said:

Can you teach your pick with axis 4 at 0 or 180 degrees?  Also, have you tried it without the optional argument Wrist?

I can’t teach the pick point with axis 4 locked. Not always at least.
no, I can try that tomorrow.

lemster68

You asked about difference between taught point being different than calculated--Not that I am aware of.  ConfL\Off--if one time the robot has axis 4 0 degrees, 5 flipped one way and 6 turned wherever it needs to be and then the next time axis 4 -180, 5 flipped the other way and 6 rotated (of course) another way, then the position will differ.