Annoying job made easy (Pin sorting automated)

[Mr Jos]

Ever been dismantling one set or more, and had to sort every pin by hand? Now you can just throw them all in 1 bin, and watch them being sorted for you!

The drum can hold over 5000 pins, but around 1000-1500 it starts to stall (but anti-stall is programmed, it will run the drum backwards and try again).

The pins try to go as fast as possible after each other, but sometimes a pin bounces a little in the swing arm making a little delay needed. It does not sort 100% correct, as it's still a 'toy'. Pins that are scanned as 'unknown' are put in the middle bin, and you can put them back in the drum to have a next try.

I have been thinking to add 1 more belt to feed these unknown pins back in the drum, to have a higher % of correct sorted, but it seems overkill.

I will start making a Stud.Io file now and once finished put it with my other machinery. At this moment I don't know what my next extraordinary machine will be. I'm open for suggestions and criticism.

Instructions available at: https://rebrickable.com/mocs/MOC-83411/Mr_Jos/technic-pin-sorting-machine-for-all-pins/#details

Edited July 30, 2021 by Mr Jos

[Hrafn]

I could use one of those, but building it would use up most of my pins!

Is the white thing halfway down the swing arm, a linear clutch of some sort?

[Gray Gear]

Very cool!

But if it only scans for size and color this doesn't work if some different but similar length pins are the same color, right?

Like black ooo and black o++ pins for example

Edited July 26, 2021 by Gray Gear

[Mr Jos]

2 minutes ago, Hrafn said:

I could use one of those, but building it would use up most of my pins!

Is the white thing halfway down the swing arm, a linear clutch of some sort?

Yes indeed, it's these 2 parts;

https://www.bricklink.com/v2/catalog/catalogitem.page?P=46835#T=C&C=85

https://www.bricklink.com/v2/catalog/catalogitem.page?P=46834#T=C&C=1

I have over 10K black 2L pins, and all others between 1K-5K I think. So still enough left to test.

2 minutes ago, Gray Gear said:

Very cool!

But if it only scans for size and color this doesn't work if some different but similar length pins are the same color, right?

Like black ooo and black o++ pins for example

Correct, the sensor only detects RedGreenBlue color reflection. Making it impossible to see the shape. Thats why I only use the o++ as it's a 'modern' pin, and the ooo pin hasn't been in a set since 2007 I think, so no problem for newer sets. (I do have a few of them but only used them in my Perforating Press to make it look good). All other pins don't really have same length with different shape (at least not 'modern' pins).

[syclone]

Impressive machine, although I've sold my NXT recently, so no pin-sorter for me 

Wonder if EV-3 has enough power to run a shape-recognition program for wider pin-sorting possibilities using a webcam. 

[howitzer]

Very nice, even if it can't differentiate every possible pin ever. I love this sort of machines, they feel "useful" even if they tend to be impractical often.

3 hours ago, Mr Jos said:

At this moment I don't know what my next extraordinary machine will be. I'm open for suggestions and criticism.

How about a machine that sorts panels by symmetry (left/right/symmetrical) in different bins?

[Mr Jos]

14 hours ago, howitzer said:

Very nice, even if it can't differentiate every possible pin ever. I love this sort of machines, they feel "useful" even if they tend to be impractical often.

How about a machine that sorts panels by symmetry (left/right/symmetrical) in different bins?

As I did make this machine I noticed why there aren't many 'pin' sorters, the little edges on the end of the pins and the ridges get stuck anywhere they can!! (That's why I added thick A4 paper to the sidewalls of the drum). First versions of this sorter looked way better and technical, but even the littlest tolerance in plates/bricks/tiles/liftarms made the pins catching and destroying the machine or the pin flying through the air.

I think this was one of the hardest types to sort as beams/axles don't have protruding edges, but I'm not going to make an other sorter soon, and I don't have enough panels to make one for that.

Progress so far in Studio.

Very much been thinking to combine 3 of my machines, but don't have enough EV3 bricks/motors yet to make it work all. Highbay warehouse crane with racks, 6Degrees of Freedom robot and the sorter. Sorting lego bricks, putting them in bins, robot picking up the bin and place it on the conveyor of the warehouse, and the crane storing it in a known place, ready to be taken out again and delivered by the 6DoF.

[Mr Jos]

Been able to model a bit faster now with Studio, made a palette with parts I use, why didn't I think of that ever before. Certainly 3x faster.

1448 Parts at the moment, checking for mistakes tomorrow, and starting with instructions.

Edited July 27, 2021 by Mr Jos

[johnnym]

@Mr Jos: Fascinating to watch. Because of the speed the parts move on the last belt, does it always work to move the arm quick enough from one extreme to the other (edge case, I know)? Or maybe you slow down the last belt, but I didn't notice a difference in the video?

If only the motors, sensors and and hub wouldn't be so expensive. Though I can't complain if I consider the amount of PF/PU motors, PF battery boxes and IR receivers and BuWizzes I have bought for my race cars. Could really use that now, as I have a few sets in parts laying around waiting to be inventorized and sorted afterwards.

[Mr Jos]

2 hours ago, johnnym said:

@Mr Jos: Fascinating to watch. Because of the speed the parts move on the last belt, does it always work to move the arm quick enough from one extreme to the other (edge case, I know)? Or maybe you slow down the last belt, but I didn't notice a difference in the video?

If only the motors, sensors and and hub wouldn't be so expensive. Though I can't complain if I consider the amount of PF/PU motors, PF battery boxes and IR receivers and BuWizzes I have bought for my race cars. Could really use that now, as I have a few sets in parts laying around waiting to be inventorized and sorted afterwards.

The last belt is always on a constant high speed (I think 1000°/sec), it never changes. The scanning belt before however I do stop in this case. You can see this very good in the beginning of the video when filming from the top on the scanning belt. When a pin is on the belt, the program knows where it will be dropped (Lets call it point 'A'). The next pin is scanned by the 2 sensors, at this moment the dropoff place is known 'B' for this 2nd pin. The program calculates the time the swingarm will need to move this amount of degrees. And it knows approx. when the 1st pin will fall off the belt+little marging. It will let out pin #2 on the last belt, and start moving it 'at the last moment'. This is done with the previous time calculated the arm will need. If all goes perfect the arm is in position right before the pin drops, next pin should be on the same belt going already if nearby, and waiting longer if a large swing is needed.

It was very fun to program this, but not easy. Right now breaking down all my MOC's to make an inventory of my parts. I already ordered a few times extra pieces I thought I ran out of but had a bag full hidden somewhere... Will take a few days I think.

For the electronics, you need patience and some luck. Yesterday I was 30seconds to late for someone giving away a complete EV3 Home Edition+few kg of Technic+studded. It was just few minutes posted on FB. But I did manage to get all my complete EV3 Edu sets there for under 200€ each, which is ok for me.