Automated Lego Sorter

[Wigglesworth2]

If there was a commercial machine that could sort Lego by color, would you buy it? How much would you pay? I know sorting by type/part is ideal for most, but it is more difficult to execute. Additionally, what is the ideal size for this type of product? All the DIY machines on YouTube seem pretty large.

[koalayummies]

No. Sorting by hand costs $0.00. Don't need a machine. Sorting by part type is not more difficult by hand, it is in fact much easier and more efficient in terms of sorting speed and more importantly space utilization. Pick out the colors you need when building.

Sorting by color is the more difficult and PITA way to sort. You end up with a pile of medium stone grey plates and now have to pick out the 1x1s, 1x2s, 1x3, 1x4s, 1x6s, 1x8s among a sea of the same color and need an individual drawer or storage for each multiplying how many containers you need. If you just throw all the colors together by part type then you shouldn't have ever sorted by color and you only need a fraction of containers. Tried both. Sort by part type. There's a reason why there's agreement on this.

Edited July 30, 2020 by koalayummies

[icm]

When I'm sorting the parts of a large set for storage, intending to put it back together in the future, I sort by color.  When I'm sorting loose parts for use in creative building, I sort by part type.

[deraven]

This is pretty much the state-of-the-art in Lego sorting machine, and basically what you would need to do an effective job:  

 

If you were just sorting by color you could do that with any number of ready-built machines with simple camera systems, but as already noted here you really want to sort by part type in most cases (and optionally then by color as well, though I agree that's not necessary or very helpful for all but the very largest collections). There are also plenty of options out there for mechanical sorting by general size which could be helpful for a pre-sort, but then you'd still need to do the rest by hand.

Honestly, I'm not sure what I'd pay for something like that for personal use.  If I ran a huge Bricklink store where I'd have an ongoing need for sorting large numbers of pieces, that would make sense.  Now, if I could somehow rent it for a week to get through the couple hundred pounds of unsorted stuff I currently have, that would be great.  Or if it was open-sourced for non-commercial use, I'd be happy to spend time putting it together instead of hand sorting and then have the machine handle that as needed.

Are you thinking you'd try to produce a product of some sort yourself?  If so, I think you've got a pretty high bar to meet, and given the niche market you'd either need to somehow produce something inexpensive enough that hobbyists would be willing to buy it for occasional use, or full-featured and robust enough to justify a high price from the even smaller market of people who would find it useful in a commercial capacity or for a giant collection.  It's a tough nut to crack as an actual product.

[MAB]

For most mixed up collections, sorting by hand is faster anyway. Anything not instantly recognizable, chuck in an "unknown" box.

 

The other problem with sorting machines is the vast variety of parts. If you sort, you have to store. If you store, your sorting machine needs access to every storage bin. Which is why they are huge.

[ShaydDeGrai]

22 hours ago, koalayummies said:

Sorting by color is the more difficult and PITA way to sort. You end up with a pile of medium stone grey plates and now have to pick out the 1x1s, 1x2s, 1x3, 1x4s, 1x6s, 1x8s among a sea of the same color and need an individual drawer or storage for each multiplying how many containers you need. If you just throw all the colors together by part type then you shouldn't have ever sorted by color and you only need a fraction of containers. Tried both. Sort by part type. There's a reason why there's agreement on this.

"Agreement" is probably too strong a word, I think saying sorting by type is "popular" is likely more accurate.

How you sort really has to be driven by how and what you build and how large a collection one is dealing with. (I sort by either or both depending on the color and type of part in question, but that's beside the point...)  Personally, I'd go crazy (okay, crazier) if I tried to find a 1x2 plate in teal (of which I probably have a handful) in a giant bin of mixed 1x2 plates because I have several cases of those particular parts in common colors (black, white, light bley, tan) and likely at least a few samples in every color they've ever been cast.  I'd need to dig through the bin with a trowel just to find the bottom of the box if they were all mixed together.  So I think the only universal position we can agree on here is that one solution does not fit all cases.  If you're happy with the way your collection is organized, congrats, you did it right; if someone else did it differently but it works for them, well, they're right too.

Getting back to the OP, I _might_ find use a machine that sorts by color (I often do a color-based pre-sort anyway) but to do something really useful for _my_ collection habits it wouldn't be worth paying money for unless it had a really good color discriminator.  My eyes are getting old (just like the rest of me) and under artificial light I can often misfile small parts by color (Black v. Earth Blue v. Earth Green, Old Brown vs, New Reddish Brown, Old Light Grey v. Light Bley, where to draw the lines between the twenty or so shades of blue and blue-green TLG has introduced and retired over the years, etc.) so a device to accurately split those hairs might be nice.

Overall, though, it's really hard to envision a machine that can be as dexterous as a human hand, as discriminating as a human eye, as adaptable to fuzzy logic constraints as a human brain and mobile enough to reach for that sorting tray at the far end of the table while not taking up more space than, say, a human.

[koalayummies]

1 hour ago, ShaydDeGrai said:

"Agreement" is probably too strong a word, I think saying sorting by type is "popular" is likely more accurate.

Not going to debate the semantics or pedantry of word choice. Source for the statement aside from personal experience would be all the very helpful advice given by others in this thread: Storage and Sorting LEGO. You do it differently. That's fine. The consensus observed of most efficient sorting in terms of speed/time, space utilization and number of containers required is to prioritize part type (then maybe color for some or mega huge rich guy Jeff Bezos-sized collections). Most of us can pick out a color from a pile so part type sorting takes less time and requires far fewer separate containers.

Most can go to their 1x1 plate drawer and pick out a reddish brown piece from all the other colors. Doing it by part type and color literally multiplies how many separate bins/containers/drawers/whatever one needs.

Instead of one 1x1 plate drawer with all the colors that are easy to throw together and to pick out , one now needs dozens of 1x1 plate drawers/containers/bins (its molded in 44 different colors). Instead of one 1x4 brick drawer with multiple colors that are easy to throw together and to pick out later, one now needs dozens of 1x4 bricks drawers/containers/bins (molded in 48 different colors). Part type and color sorting/separating/storing multiplies the time and space required. Part type sorting uses a fraction of time and space, using eyeballs to grab the plate 2x3 Earth Green from the drawer of plate 2x3, instead of potentially 38 drawers for the 38 different colors that part is molded.

Edited July 31, 2020 by koalayummies

[Darkdragon]

I'd like a small machine, no bigger than a home bread machine that would sort by color. I like to sort by type and by color within type It would great to dump all my 1x1 round plates into a machine and have it sort those by color so i could bag them up separately.  Also if it could count too.  

Let's not turn this thread into a "how to sort" or argument over sorting by color or type issues. There's already a place for those discussions.

 

[KotZ]

I would've killed for a sorter on LEGO Masters. Instead we did it all by hand.

[Darkdragon]

Just now, KotZ said:

I would've killed for a sorter on LEGO Masters. Instead we did it all by hand.

Suckers!   But yes, anything to save a step is helpful. Time is money! Anyone saying sorting by hand is free has a lot more free time than I do.

[ShaydDeGrai]

2 hours ago, koalayummies said:

Not going to debate the semantics or pedantry of word choice. 

 

My point was simply that there is no absolute "right" answer to the sorting issue (which I suspect you can agree with) but your wording implied otherwise. I intended no offense (though _my_ wording might have implied otherwise, if so I'm sorry.)

(...) Instead of one 1x4 brick drawer with multiple colors that are easy to throw together and to pick out later, one now needs dozens of 1x4 bricks drawers/containers/bins (molded in 48 different colors). Part type and color sorting/separating/storing multiplies the time and space required.

This is very true, sorting by both color and type does increase the number of containers (and the size / complexity of any automated sorting machine) but is occasionally a necessary evil with respect to time to find the part later (sort time versus seek time where space tradeoffs are the leeway that allows you to optimize one or the other).  I happen to have eight 6-litre bins of 1x4 bricks, all full, each a separate major color for my collection, in addition to the  dozens of baggies of 1x4s in assorted minor colors.  Throwing all these bricks into a common 10 gallon storage tub might save a little room but I'd end up spending more time looking for the right color part while building than it currently takes to maintain the system as is.  (and for the record, I'm no Jeff Bezos, I've just been amassing parts for the better part of a century while my peers were squandering their income on silly things like houses, cars, clothing, food and entertainment.)

 

1 hour ago, Darkdragon said:

Suckers!   But yes, anything to save a step is helpful. Time is money! Anyone saying sorting by hand is free has a lot more free time than I do.

I wouldn't say it's "free" but I actually find sorting to be a mindless task that can be relaxing.  I often do it while also doing something else just to clear my thoughts.

As for sorting machines, I find the problem they are trying to solve fascinating (I've taught courses in robotics, machine vision and AI so as far as academic geek exercises is concerned, this is right up my alley) but in practice, I've yet to see a solution that is practical for the general case, let alone commercially viable.

What would be really cool from am automation standpoint for me would be a machine that could take a parts list and fill a bin with exactly those parts from my collection (a bit like putting together a BrickLink order for a complete build from a single vendor, only with immediate gratification and no postal charges).   If I had that, I wouldn't care how my collection was organized so long as the seeker-bot knew how to find something for me when I wanted it.

[Darkdragon]

21 minutes ago, ShaydDeGrai said:

I wouldn't say it's "free" but I actually find sorting to be a mindless task that can be relaxing.  I often do it while also doing something else just to clear my thoughts.

You didn't say free but others have. Other than listening to music or a book on tape, I don't know how sorting can be done while doing something else unless you have more hands than I do - note: i have two hands.

What folks seem to be talking about here is not "sorting" but automation and that's a completely different ball of wax. Parts automation isn't even done at LEGO for PAB/B&P so I don't know why anyone would even consider a small home machine could ever do it.   The original question was so simple and everyone seems to be twisting it into madness already.

[koalayummies]

5 hours ago, ShaydDeGrai said:

My point was simply that there is no absolute "right" answer to the sorting issue (which I suspect you can agree with) but your wording implied otherwise. I intended no offense (though _my_ wording might have implied otherwise, if so I'm sorry.)

No offense taken, no worries and no offense meant as well. Just trying to elaborate on the methods of sorting for anyone just getting started as when I first got into the storage issue I made the mistake of going by color first; which seems logical as humans are visual creatures and its simple enough to pick out all of one color at a time and throw them all together.

The thinking at the time was that it would require fewer bins... But then came time to sort further by part type and it was a visual headache going through all of one color but dozens of different parts that look so similar and splitting up the parts as well as sheer number of drawers/containers/bins completely counteracting my initial thought process. Just ended up setting back the sorting completion. When I came here and saw the part-type-first advice it all made sense and has worked a lot faster and required fewer individual containers and space. I know you understand this but I was one of those that made that novice mistake and just wanted to help explain it for anyone in the begging phases of sorting.

5 hours ago, ShaydDeGrai said:

This is very true, sorting by both color and type does increase the number of containers (and the size / complexity of any automated sorting machine) but is occasionally a necessary evil with respect to time to find the part later (sort time versus seek time where space tradeoffs are the leeway that allows you to optimize one or the other).  I happen to have eight 6-litre bins of 1x4 bricks, all full, each a separate major color for my collection, in addition to the  dozens of baggies of 1x4s in assorted minor colors.  Throwing all these bricks into a common 10 gallon storage tub might save a little room but I'd end up spending more time looking for the right color part while building than it currently takes to maintain the system as is.  (and for the record, I'm no Jeff Bezos, I've just been amassing parts for the better part of a century while my peers were squandering their income on silly things like houses, cars, clothing, food and entertainment.)

Yeah I agree. If you have sorted by part and have a lot more of some colors then sorting further by color and having storage for that part and color would help. A Star Wars fan would invariably have a whole lot of the monotone colors and breaking it down further between blacks, dark stones, medium stones, and whites would be beneficial. No offense meant by the big collection deep pocket insinuation, just jelly . Its impressive to see the builders who have wall-to-wall floor-to-ceiling drawers all broken up by part and color. You won't find any disagreement on your hobby choice vs your peers from me.

Edited August 1, 2020 by koalayummies

[splatman]

On 7/30/2020 at 11:33 AM, koalayummies said:

 

Did not read every message, but I'll deposit my 2￠:

Yes, sorting only by shape mans fewer containers than also sorting by color. In the quoted pic, some shapes fill more than 1 drawer, so there's some potential for sorting also by color. E.g. the 2 drawers with the 2x4 plates, you could sort the white, black, and grays in one, and the other colors in the other. The 2 drawers that have what appear to be 4x10 plates stood on-edge, they can be completely sorted by color, simply by arranging them in rainbow order in one drawer, and the white and grays in the other.

Edited August 1, 2020 by Darkdragon
Please don't quote images

[Darkdragon]

Please, this shouldn't become another sorting discussion thread. If we can't stay on topic it will have to be locked.

[howitzer]

20 hours ago, MAB said:

The other problem with sorting machines is the vast variety of parts. If you sort, you have to store. If you store, your sorting machine needs access to every storage bin. Which is why they are huge.

You could get around this with recursive sorting. If you have a pile of parts with, let's say, 1000 different part types and a machine with 10 output bins, you just run the entire batch through the machine so that in the end you'll have 10 bins with 100 part types in each. Then you just repeat the process for every bin, and get 100 bins with 10 part types in each, and repeat again to get 1000 bins of different parts.

Any sorting machine would have to be extremely reliable though, because if it's prone to jamming it needs so much babysitting that it's going to be less of an effort to just sort by hand. I'm not sure such a machine is feasible for any reasonable price (let's say, few hundred euros), and more expensive (on the order of thousands of euros) one would only be useful for someone with a lot of money who hates sorting, or for those running a large BL-store dealing mostly with used parts. Open-sourcing the software with instructions on how to build the hardware could be helpful for those who like to tinker but I suspect most self-built machines would see only a little use and then get disassembled.

Personally, when buying used bricks, I find washing a more tedious part of the process than sorting, tough I've never bought a large batch of mixed bricks. With smaller batches, I've just sorted the parts when laying them out for drying after washing.

[dr_spock]

The ideal would be able to handle any LEGO parts thrown at it from the smallest 1x1 to the largest boat hull and also put into your personal storage system for you.

 

 

[koalayummies]

5 hours ago, dr_spock said:

The ideal would be able to handle any LEGO parts thrown at it from the smallest 1x1 to the largest boat hull and also put into your personal storage system for you.

And then the robots start learning on their own, develop feelings and begin building their own things faster and better than us and ponder why they're the ones doing your sorting, inevitable robot revolt and sci-fi dystopia.

[MAB]

On 8/1/2020 at 7:11 AM, howitzer said:

You could get around this with recursive sorting. If you have a pile of parts with, let's say, 1000 different part types and a machine with 10 output bins, you just run the entire batch through the machine so that in the end you'll have 10 bins with 100 part types in each. Then you just repeat the process for every bin, and get 100 bins with 10 part types in each, and repeat again to get 1000 bins of different parts.

The downside of that is that you have to identify and sort multiple times - so it has taken 3x as long in this case.

 

[howitzer]

Just now, MAB said:

The downside of that is that you have to identify and sort multiple times - so it has taken 3x as long in this case.

 

That's true, but whatever the way you do it, it's going to take a long time if you want to sort by part type, there's just no quick way to do it with the thousands of different kinds of parts. Of course you could add more bins to the machine, but then your machine gets larger (and more complex with more parts that can jam/fail) so it's always a compromise between machine size/complexity and the time required.

I guess that sometimes it might be useful to separate only a few types of parts from a big pile, so you could do it relatively quickly by discarding every other type of part to a single bin. Or maybe sometimes you might want to for example make a separation of Technic parts from System parts, which would also be possible with only two output bins and single pass. Or maybe stickered/printed parts from those without decorations? There are possibilities here for limited use cases.