MoveAbsJ/MoveJ joint interpolation

Ilian Bonev

Hello everyone,

 

Let, for example,

var jointtarget j1:=[[-90,  0,-60,  0,-30,  0],[9E9,9E9,9E9,9E9,9E9,9E9]];
var jointtarget j2:=[[-80,10,-50,10,-20,10],[9E9,9E9,9E9,9E9,9E9,9E9]];

var robtarget r1;
var robtarget r2;

...
r1:=CalcRobT(j1,tool0);

r2:=CalcRobT(j2,tool0);
...

 

Let the robot be at j1 (i.e., at r1).

 

What is the profile of the six joint values during the execution of

MoveAbsJ j2,v100,fine,tool0;

or

MoveJ r2,v100,fine,tool0;

 

I know that in both cases, all joints start and finish at the same time, so I always thought that after acceleration and prior to deceleration, the profile for each joint is linear, but I was surprised to see that it is not. In fact, in some cases, a joint that has to go from say 10 to 30 degrees, will first go to say -5 degrees and then to 30 degrees.

 

Does anyone know details on the actual trajectory planning? I want to be able to foresee the path of a robot in a MoveJ or MoveAbsJ motion. Is this some kind of an approximation of the minimum-time trajectory?

 

Thanks,

 

Ilian

 

lemster68

Greetings,

 

Is it by chance that the robot wrist is at some point directly above the center of axis one?  This is an additional robot singularity, I have seen axis one swing around quite a bit with a MoveJ instruction in these cases.  A MoveAbsJ cures that.

Ilian Bonev

Hi,

 

I did some further tests and I realized that I was wrong. Besides, it is clearly explained in the ABB manuals that the central portion of the joint-space trajectory of any joint is linear, during a MoveJ (or MoveAbsJ) motion. And you are right -- there are exceptions when near singularities.

 

By the way, you are referring to what is called a "shoulder singularity," which in the popular inline wrist six-axis robot occurs when the distance between the center of the robot and the axis of joint 1 is minimum. See slide 27 of my lecture notes: http://www.gpa.etsmtl.ca/cours/gpa546/Notes/Cours11.pdf

 

Cheers,

 

Ilian