Aloha Forum, Hi Juha,
First, allow me to say: This thing rocks! To be honest, I'm tempted ordering one for my hobby shop. Not to complain, the placing speed could be better and I would like to share my ideas.
#1 Measure the distance between 2 "landing holes" (sorry, i don't know that they are exactly called) in the component stripe.
I assume you know "the offset" between camera and pickup tool; you could use this "two hole distance" to estimate the next position for pickup and do position calibration "almost in the right spot".
This could/would eliminate/improve the "calibrate and move head" step seen in the video.
I'm not sure what happens/how it's handled, should the "stripe landing holes" run out; for example i've placed the exact amount of parts needed on the table and the last parts needs to be picked up...
#2 Mount the cameras left and right of the pickup tool.
This would unfortunately limit the parts "placeability" to "from top only" (which wouldn't bother me when using "ebay feeders"). However, with the 2 cameras you could always "land on the right spot"; should the position verification fail from the right camera, use the left or vice versa.
#3 Can you detect should a pickup fail, or the part is dropped before placement (pump/hose failed, etc)? A pressure sensor (even a 2 state simple one) could solve this.
Hope these will help you improve your already great design.
Cheers, Peter
Speed up ideas and stuff.
Re: Speed up ideas and stuff.
Hi Peter, welcome to the forum!
Yes, the current way of looking for each hole separately is kind of dumb. In near future it will look for the first and last holes only. Of course, these do give the location of the holes in between.
You need to cut your tapes so, that the hole defining the location of a part goes with the part.
I don't detect a dropped part, nor have I found this to be a problem. It would be useful to notice that there wasn't a part in the first place; now, you have to be careful when removing the cover tape. The tapes are held in place by the double sided tape and this makes it easy, but you still need to keep the parts in their places.
Yes, the current way of looking for each hole separately is kind of dumb. In near future it will look for the first and last holes only. Of course, these do give the location of the holes in between.
You need to cut your tapes so, that the hole defining the location of a part goes with the part.
I don't detect a dropped part, nor have I found this to be a problem. It would be useful to notice that there wasn't a part in the first place; now, you have to be careful when removing the cover tape. The tapes are held in place by the double sided tape and this makes it easy, but you still need to keep the parts in their places.
Re: Speed up ideas and stuff.
Hi Juku.
I think we've misundertstood each other..
I'll try to explain again what i meant with #2.
It seems to me, that with the current camera setup, when you pick "one of the last" parts, the camera is "off the stripe". In this case you cant reliably verify your position, should you "return to the next estimated position", as I written in #1 (pickup head is above the part, camera is "of the stripe").
With a 2 cameras "on the sides", should this happen, you can still verify your position with the second one. When moving left to right, sooner or later cant the right camera verify the position. The left will however be above the stripe and can provide verification...
Cheers, Peter
I think we've misundertstood each other..
I'll try to explain again what i meant with #2.
It seems to me, that with the current camera setup, when you pick "one of the last" parts, the camera is "off the stripe". In this case you cant reliably verify your position, should you "return to the next estimated position", as I written in #1 (pickup head is above the part, camera is "of the stripe").
With a 2 cameras "on the sides", should this happen, you can still verify your position with the second one. When moving left to right, sooner or later cant the right camera verify the position. The left will however be above the stripe and can provide verification...
Cheers, Peter
Re: Speed up ideas and stuff.
> with the current camera setup, when you pick "one of the last" parts, the camera is "off the stripe".
(I think that the stripe here refers to the partss tape.)
The machine knows very well the offset of the camera and the needle, therefore the above does not really happen. Assuming you cut the tape correctly (that is, the hole defining the last part location is there), what happens with the last part is that the camera goes and measures the last hole. From this measurement, the system knows the location of the part. Then, the needle is taken to the part. At this point, the camera is indeed off the tape but this doesn't matter, as the needle location is still known. The system picks up the part and takes it where it belongs. Ok?
(I think that the stripe here refers to the partss tape.)
The machine knows very well the offset of the camera and the needle, therefore the above does not really happen. Assuming you cut the tape correctly (that is, the hole defining the last part location is there), what happens with the last part is that the camera goes and measures the last hole. From this measurement, the system knows the location of the part. Then, the needle is taken to the part. At this point, the camera is indeed off the tape but this doesn't matter, as the needle location is still known. The system picks up the part and takes it where it belongs. Ok?
Re: Speed up ideas and stuff.
Not important, but just because I didn't understand what this was really about (and in case nobody does - haha) :
As how it works at the moment, it seems to be not consistent. And it it would, the subject wouldn't be there (I guess).
Current working :
1. Find strip and first component (with camera).
2. Don't ask me what the "calibrating" exactly is about, because the camera can't see the needle. So this should be about the detailed positioning of a hole further down the strip by vision, knowing that the needle is at a fixed position from the camera and hole distance is also fixed.
3. At the next round of pick, the camera seems to look for the the next component and when found the head moves to it and picks it up.
4. When the strip runs out for the last couple of components the camera again first looks for the component and moves the head.
The latter (#4) is very vaguely written by me but this is because of this :
Curiously, #3 seeems to be there to serve #4; Standard way of working. So the looking ahead of the camera should be able to perform the very same in order to let all work from start to end (of the strip). And I think the text in the quote denotes a special "end procedure".
Then how it could work :
1. Find strip and first component (with camera) and calibrate as needed.
2. Find last component in that strip as per quantity needed for this batch (needed can be 20, strip can contain 50). It it assumed that no components are lacking in between (user responsibility).
3. Race to the first component and pick it.
4. At the next round, race directly to the second component and pick it.
Looks simple enough.
But maybe it isn't ...
(even though Juha - I think - proposes the latter himself; I will create a new topic with possible "issues" around this)
Peter
As how it works at the moment, it seems to be not consistent. And it it would, the subject wouldn't be there (I guess).
Current working :
1. Find strip and first component (with camera).
2. Don't ask me what the "calibrating" exactly is about, because the camera can't see the needle. So this should be about the detailed positioning of a hole further down the strip by vision, knowing that the needle is at a fixed position from the camera and hole distance is also fixed.
3. At the next round of pick, the camera seems to look for the the next component and when found the head moves to it and picks it up.
4. When the strip runs out for the last couple of components the camera again first looks for the component and moves the head.
The latter (#4) is very vaguely written by me but this is because of this :
With the notice that this is not on the video.what happens with the last part is that the camera goes and measures the last hole. From this measurement, the system knows the location of the part.
Curiously, #3 seeems to be there to serve #4; Standard way of working. So the looking ahead of the camera should be able to perform the very same in order to let all work from start to end (of the strip). And I think the text in the quote denotes a special "end procedure".
Then how it could work :
1. Find strip and first component (with camera) and calibrate as needed.
2. Find last component in that strip as per quantity needed for this batch (needed can be 20, strip can contain 50). It it assumed that no components are lacking in between (user responsibility).
3. Race to the first component and pick it.
4. At the next round, race directly to the second component and pick it.
Looks simple enough.
But maybe it isn't ...
(even though Juha - I think - proposes the latter himself; I will create a new topic with possible "issues" around this)
Peter
Re: Speed up ideas and stuff.
So here : Re-initialisationLooks simple enough.
But maybe it isn't ...
(even though Juha - I think - proposes the latter himself; I will create a new topic with possible "issues" around this)