Path planning with industrial robots with tracking

On Jan 16, 12:16 pm, e1e120032...@gmail.com wrote:

Can't you just predict where end point is going to be and move the
manipulator there?  A more complicated way is to have a formula the
moves the manipulator as a function of the converyor position and then
use the chain rule to adjust for the conveyor speeds.


Why not get a motion controller that can do this.

Peter Nachtwey

What you wrote is not path planning. I need function p(t9 (t is time
and p is distance from beginnig point).
Every 1msec I write this position to servo controller and servo
controller changes speed, acceleration etc.
It means this p(t) is not known at the start of the moving (because
endpoint is not fix). If endpoint is fix, I have solution.

Best regards

Leo

Sure it is.  It predicts where the point 2 is will be and goes there.


That is wasteful.  The motion controller better be able to change the
motion profile smoothly on-the-fly to do that.


I think what you are asking for is a superimposed move.   In this case
you tell the motion controller to move the maniplutor to the current
point 2.  You said you can do that.   At the same time you add the
motion of point to the motion of the manipilator.  You are adding two
motion profiles together.  When the first motion profile is done the
manipulator will be at point 2 where ever that is on the converyor.
Simple.   The downside of this technique is that the manipulator will
mimick all the uneven motion from the conveyor.

Peter Nachtwey

TRhere is only one motion profile.

OK, It is not clear what you mean.  How can ONE motion profile move
between point 1 and point 2 when point 2 can be in more that one place
dpending on the conveyor speed?  Some part of the target generator
must take into account the motion of point 2 on the manipulator.  If
you aren't going to put the effort into explaining the problem then
you aren't going to get the answers you need.en

Peter Nachtwey

OK, Peter. I have algorithm if point 1 is fix and point 2 is fix. It
means, I have coordinate for point 1(X1,Y1,Z1)
and I have coordinate for point 2 (X2,Y2,Z2). Now, I know distance
between between this two points and I can calculate
trajectory between points P1 and P2 as function of time:P(t) (as input
values are max.velocity,max.acceleration,max.jerk...).Next,
every 1 msec I write new value in servo drives( if I have 3 drives,
then I have to calculate move for every drive):P(t1)P(t2)......P(tn)
.P(t1) is pont(X1,Y1,Z1) and point P(tn) is P(X2,Y2,Z2). My problem
is:Which algorith can I use if point P(X2,Y2,Z2)is not fix, but I know
every moment
what is coordinate of this point.
Regards

Leo

Leo, I don't think you have understood what I am saying about super-
imposed moves.  The two motion profiles are run concurrently, not
sequentially.
That is pvaj(t)=pvaj1(t)+pvaj2(t).   pvaj is my term for position,
velocity, acceleration and jerk.   pvaj1(t) is the motion profile to
point 2 at time 0.
pvaj2(t) is derived from the actual motion of the conveyor.  Every
millisecond the pvaj1(t) and pvaj2(t) are calculated and summed yield
pvaj(t). pvaj(t) is the target that the manipulator will follow.
Note, is is often hard to estimate the acceleration and jerk from the
conveyor with out the use of an Kalman filter or observer.

Peter Nachtwey