The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
Happy sorting,
Matt
Is this an ad for a product? People have been promoting AI/algorithmic part sorting
for at least 5 years.
Hi -- my apologies for not being clear. This sorter goes way beyond any
of those object identifiers you mention. Using industrial robotics, computer
vision, and machine learning, this sorter actually carries out the sorting task
for you. Demo videos and more info available upon request -- please PM me.
Cheers,
Matt
In General, Andy_Bell writes:
In General, mcheselka writes:
The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
Happy sorting,
Matt
Is this an ad for a product? People have been promoting AI/algorithmic part sorting
for at least 5 years.
The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
Happy sorting,
Matt
I'm in no way telling you how to run your business, but in this day and age
why not post a youtube video or two and put up a link? Not too many people read
this forum unfortunately.
The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
Happy sorting,
Matt
If it can open boxes and bags when parting out new sets, I'm in! That's
the bit that takes the most amount of time.
Needs to be done neatly though - opening them carefully so they can be re-used
for sending orders. Sorted by size of box/bag, and stacked neatly please...
Used parts - great if it could separate ALL parts - including 3/4 pins still
stuck in parts with a hole). And wash them before they go to the Automating Sorting
Assistant .
Oh. At least 500kgs at a time please. Otherwise we can most likely do it quicker
by (almost) hand.
"if and when" - I think an Australian fella did what you mention a few
years ago. It's out there on YouTube already...
LOL not this one but probably in a future version...
Cheers,
Matt
In General, MyTwoPence writes:
In General, mcheselka writes:
The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
Happy sorting,
Matt
If it can open boxes and bags when parting out new sets, I'm in! That's
the bit that takes the most amount of time.
Needs to be done neatly though - opening them carefully so they can be re-used
for sending orders. Sorted by size of box/bag, and stacked neatly please...
Used parts - great if it could separate ALL parts - including 3/4 pins still
stuck in parts with a hole). And wash them before they go to the Automating Sorting
Assistant .
Oh. At least 500kgs at a time please. Otherwise we can most likely do it quicker
by (almost) hand.
"if and when" - I think an Australian fella did what you mention a few
years ago. It's out there on YouTube already...
If it can open boxes and bags when parting out new sets, I'm in! That's
the bit that takes the most amount of time.
Needs to be done neatly though - opening them carefully so they can be re-used
for sending orders. Sorted by size of box/bag, and stacked neatly please...
Used parts - great if it could separate ALL parts - including 3/4 pins still
stuck in parts with a hole). And wash them before they go to the Automating Sorting
Assistant .
Oh. At least 500kgs at a time please. Otherwise we can most likely do it quicker
by (almost) hand.
"if and when" - I think an Australian fella did what you mention a few
years ago. It's out there on YouTube already...
... just MyTwoPence Bricks
I was thinking the same thing. Recognizing a piece is easy with today's
technology. But separating the pieces is more complex. Once I separate the
pieces they are already in my hand and I know what the piece is. But I have
thought that it would be interesting to build a sorter out of LEGO more just
for the challenge of doing it not as a practical tool for running a store. But
all of the designs in my head require a camera.
If it can open boxes and bags when parting out new sets, I'm in! That's
the bit that takes the most amount of time.
Needs to be done neatly though - opening them carefully so they can be re-used
for sending orders. Sorted by size of box/bag, and stacked neatly please...
Used parts - great if it could separate ALL parts - including 3/4 pins still
stuck in parts with a hole). And wash them before they go to the Automating Sorting
Assistant .
Oh. At least 500kgs at a time please. Otherwise we can most likely do it quicker
by (almost) hand.
"if and when" - I think an Australian fella did what you mention a few
years ago. It's out there on YouTube already...
... just MyTwoPence Bricks
I was thinking the same thing. Recognizing a piece is easy with today's
technology. But separating the pieces is more complex. Once I separate the
pieces they are already in my hand and I know what the piece is. But I have
thought that it would be interesting to build a sorter out of LEGO more just
for the challenge of doing it not as a practical tool for running a store. But
all of the designs in my head require a camera.
LEGO MOC's do exist for sorting various types of brick or plate or liftarm,
and some can even do more complex pieces. But it takes a lot of LEGO to build
a sorter and some fine-tuning to properly sort out the different elements.
Google or your preferred search engine is ideal for finding these LEGO MOC's.
They would be pretty large if they could sort all part types. Larger than an
average LEGO collection I would imagine.
The biggest hurdle is sorting the pieces without hurting them during sorting.
A lot of MOC's I've seen over the years prefer to avoid harming the bricks.
Some MOC's can even automate assembly of parts that should be assembled,
to make repeating patterns easier to put together. It's an amazing collection
out there, good luck finding what you're after!
The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
Happy sorting,
Matt
Out of curiosity could you post some rough cost estimates? Also would it be good
for example if I parted out say 100x sets and dumped all of 1 bag in each time
would it be efficient at sorting that in a “fast” amount of time? Also like the
others mentioned if you post demo videos on yt or such then you may have interest
from people
posting links to yt videos or other business-related info/sites is against the
rules of this public forum as far as I can tell -- and therefore out of courtesy
and to keep my account active, I won't. If you're interested in the
sorter please PM me and I'll be happy to answer your questions and send you
links to demo videos.
Cheers,
M
In General, Nubs_Select writes:
In General, mcheselka writes:
The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
Happy sorting,
Matt
Out of curiosity could you post some rough cost estimates? Also would it be good
for example if I parted out say 100x sets and dumped all of 1 bag in each time
would it be efficient at sorting that in a “fast” amount of time? Also like the
others mentioned if you post demo videos on yt or such then you may have interest
from people
I knew it! I tapped on the link and immediately began to think, “what would he
post” and then I was like “surely he wouldn’t rickroll us!” But alas 😭😭😭
I knew it! I tapped on the link and immediately began to think, “what would he
post” and then I was like “surely he wouldn’t rickroll us!” But alas 😭😭😭
I thought about something cleverer… but, in the end, I couldn’t not do it the
old true way
I knew it! I tapped on the link and immediately began to think, “what would he
post” and then I was like “surely he wouldn’t rickroll us!” But alas 😭😭😭
I thought about something cleverer… but, in the end, I couldn’t not do it the
old true way
I knew it! I tapped on the link and immediately began to think, “what would he
post” and then I was like “surely he wouldn’t rickroll us!” But alas 😭😭😭
I knew it! I tapped on the link and immediately began to think, “what would he
post” and then I was like “surely he wouldn’t rickroll us!” But alas 😭😭😭
Nooooooo!!!!
This is ridiculous!!!
I though the forum was free from that stuff😭😭😭😭
Thankfully I clicked off as soon as I saw the video starting up😅
I say, this is the only reason why that video has so many views
Us LEGO fanatics should have our own video to trick people with
Thanks SylvainLS for the clarification and the great link -- I've always
dreamed of being Rick-bombed!
Cheers,
Matt
Thank you for the video`s, at first I thought it was a scam, but luckely not.
Seems to be a solid demo for a trainee, but not for in a production environment.
It is way to slow, and if it only does 90% accuracy. I mean the averge 14 year
old does 2 kg per hour, and probably above 98% accuracy.
Usually a industrial machines like a lego sorter costs above $100k so before
you got that back with 14 year olds paying them minimum wage.
Thanks SylvainLS for the clarification and the great link -- I've always
dreamed of being Rick-bombed!
Cheers,
Matt
Thank you for the video`s, at first I thought it was a scam, but luckely not.
Seems to be a solid demo for a trainee, but not for in a production environment.
It is way to slow, and if it only does 90% accuracy. I mean the averge 14 year
old does 2 kg per hour, and probably above 98% accuracy.
Usually a industrial machines like a lego sorter costs above $100k so before
you got that back with 14 year olds paying them minimum wage.
Thanks SylvainLS for the clarification and the great link -- I've always
dreamed of being Rick-bombed!
Cheers,
Matt
Thank you for the video`s, at first I thought it was a scam, but luckely not.
Seems to be a solid demo for a trainee, but not for in a production environment.
It is way to slow, and if it only does 90% accuracy. I mean the averge 14 year
old does 2 kg per hour, and probably above 98% accuracy.
Usually a industrial machines like a lego sorter costs above $100k so before
you got that back with 14 year olds paying them minimum wage.
LOL Yes!!! This one is for real. I've solved a problem that's been
around for waaaay too long!
I don't have a finished BoM yet, but the cost layout looks like it'll
be about US $5,000. Most of that is the price of the industrial robotic arm.
You buy all the materials from their sources (I provide a complete list in the
BoM), and assemble it with detailed instructions and support you'll get from
me. There will also be a very low monthly/annual license fee (TBD) to run the
software and be a part of the network.
To help you decide if the ROI is acceptable, I've made a simple calculator:
What this sorter lacks in speed it makes up for in time, since it's designed
to operate unattended 24/7.
I appreciate the feedback,
Matt
In General, Mr_Rofl writes:
Thank you for the video`s, at first I thought it was a scam, but luckely not.
Seems to be a solid demo for a trainee, but not for in a production environment.
It is way to slow, and if it only does 90% accuracy. I mean the averge 14 year
old does 2 kg per hour, and probably above 98% accuracy.
Usually a industrial machines like a lego sorter costs above $100k so before
you got that back with 14 year olds paying them minimum wage.
Is there an estimation about how much it does in 24 hours? How are the maintenance
intervals? The first machine without downtime has to be invented yet. I highly
doubt it will ever be.
In General, mcheselka writes:
LOL Yes!!! This one is for real. I've solved a problem that's been
around for waaaay too long!
I don't have a finished BoM yet, but the cost layout looks like it'll
be about US $5,000. Most of that is the price of the industrial robotic arm.
You buy all the materials from their sources (I provide a complete list in the
BoM), and assemble it with detailed instructions and support you'll get from
me. There will also be a very low monthly/annual license fee (TBD) to run the
software and be a part of the network.
To help you decide if the ROI is acceptable, I've made a simple calculator:
What this sorter lacks in speed it makes up for in time, since it's designed
to operate unattended 24/7.
I appreciate the feedback,
Matt
In General, Mr_Rofl writes:
Thank you for the video`s, at first I thought it was a scam, but luckely not.
Seems to be a solid demo for a trainee, but not for in a production environment.
It is way to slow, and if it only does 90% accuracy. I mean the averge 14 year
old does 2 kg per hour, and probably above 98% accuracy.
Usually a industrial machines like a lego sorter costs above $100k so before
you got that back with 14 year olds paying them minimum wage.
Is there an estimation about how much it does in 24 hours? How are the maintenance
intervals? The first machine without downtime has to be invented yet. I highly
doubt it will ever be.
it says
"There are about 227 objects per pound
Current Automated Sorting Rate: 1.42 pounds per hour"
on the page which would be 5448 bricks/day or 34.08 lbs/day
Is there an estimation about how much it does in 24 hours? How are the maintenance
intervals? The first machine without downtime has to be invented yet. I highly
doubt it will ever be.
it says
"There are about 227 objects per pound
Current Automated Sorting Rate: 1.42 pounds per hour"
on the page which would be 5448 bricks/day or 34.08 lbs/day
Oh, I just did 84,200 seconds per day divided by 7 seconds per piece for 12,342.
Correct, and this is as slow as you'll ever see it run. In all of my tests
sofar including the demo videos, the arm is operating at about 1/2 it's max
speed, and there's a LOT of telemetry/diagnostics being written to the computer's
HD which also slows things down substantially. There are also more efficient
ways of moving, but that'll have to wait for later versions. Having said
all that -- it will probably always be slower than a human sorter if you're
comparing rates rather than volumes.
Also keep in mind this sorter is intended to *assist*, not completely replace,
human sorters. It can't do "everything" -- yet. This will lower
your labor costs and time quite a bit, but it definitely won't eliminate
them. I know there's much more to a store than just sorting, but I'm
hoping that some assistance will free you up to focus on the more important things.
Cheers,
Matt
In General, Nubs_Select writes:
In General, Mr_Rofl writes:
Is there an estimation about how much it does in 24 hours? How are the maintenance
intervals? The first machine without downtime has to be invented yet. I highly
doubt it will ever be.
it says
"There are about 227 objects per pound
Current Automated Sorting Rate: 1.42 pounds per hour"
on the page which would be 5448 bricks/day or 34.08 lbs/day
Correct, and this is as slow as you'll ever see it run. In all of my tests
sofar including the demo videos, the arm is operating at about 1/2 it's max
speed, and there's a LOT of telemetry/diagnostics being written to the computer's
HD which also slows things down substantially. There are also more efficient
ways of moving, but that'll have to wait for later versions. Having said
all that -- it will probably always be slower than a human sorter if you're
comparing rates rather than volumes.
Also keep in mind this sorter is intended to *assist*, not completely replace,
human sorters. It can't do "everything" -- yet. This will lower
your labor costs and time quite a bit, but it definitely won't eliminate
them. I know there's much more to a store than just sorting, but I'm
hoping that some assistance will free you up to focus on the more important things.
Cheers,
Matt
indeed, even if it isn't as fast as a human sorter it could still be worth
it. if your able to make it work for new set partouts in some way and able to
increase its speed then even at $5k that would be something I can tell you 100%
as a business i would be interested in. so if your able to improve it over time
then I would love to buy something like that in the future. the thought of being
able to dump say 100 copies of a bag into a container and then leave it for a
day or so and then its all sorted while i work on other things for the store
would be amazing and worth a lot of money. (perhaps even something that could
speed it up for a mode for new sets is it always checks to see if the part its
figuring out is a part its done during that session before it checks the others
since for new set bags it would usually be like 25-50 or less unique lots per
bag)
Correct, and this is as slow as you'll ever see it run. In all of my tests
sofar including the demo videos, the arm is operating at about 1/2 it's max
speed, and there's a LOT of telemetry/diagnostics being written to the computer's
HD which also slows things down substantially. There are also more efficient
ways of moving, but that'll have to wait for later versions. Having said
all that -- it will probably always be slower than a human sorter if you're
comparing rates rather than volumes.
Also keep in mind this sorter is intended to *assist*, not completely replace,
human sorters. It can't do "everything" -- yet. This will lower
your labor costs and time quite a bit, but it definitely won't eliminate
them. I know there's much more to a store than just sorting, but I'm
hoping that some assistance will free you up to focus on the more important things.
Cheers,
Matt
indeed, even if it isn't as fast as a human sorter it could still be worth
it. if your able to make it work for new set partouts in some way and able to
increase its speed then even at $5k that would be something I can tell you 100%
as a business i would be interested in. so if your able to improve it over time
then I would love to buy something like that in the future. the thought of being
able to dump say 100 copies of a bag into a container and then leave it for a
day or so and then its all sorted while i work on other things for the store
would be amazing and worth a lot of money. (perhaps even something that could
speed it up for a mode for new sets is it always checks to see if the part its
figuring out is a part its done during that session before it checks the others
since for new set bags it would usually be like 25-50 or less unique lots per
bag)
Could definitely be worth it for smaller sets such as
Is there an estimation about how much it does in 24 hours? How are the maintenance
intervals? The first machine without downtime has to be invented yet. I highly
doubt it will ever be.
The second law of thermodynamics demands that pain and suffering be mandatory
whenever you try to make something easy
In General, mcheselka writes:
LOL Yes!!! This one is for real. I've solved a problem that's been
around for waaaay too long!
I don't have a finished BoM yet, but the cost layout looks like it'll
be about US $5,000. Most of that is the price of the industrial robotic arm.
You buy all the materials from their sources (I provide a complete list in the
BoM), and assemble it with detailed instructions and support you'll get from
me. There will also be a very low monthly/annual license fee (TBD) to run the
software and be a part of the network.
To help you decide if the ROI is acceptable, I've made a simple calculator:
What this sorter lacks in speed it makes up for in time, since it's designed
to operate unattended 24/7.
I appreciate the feedback,
Matt
In General, Mr_Rofl writes:
Thank you for the video`s, at first I thought it was a scam, but luckely not.
Seems to be a solid demo for a trainee, but not for in a production environment.
It is way to slow, and if it only does 90% accuracy. I mean the averge 14 year
old does 2 kg per hour, and probably above 98% accuracy.
Usually a industrial machines like a lego sorter costs above $100k so before
you got that back with 14 year olds paying them minimum wage.
First off; I applaud anyone who's trying to perfect this for end users rather
then having us ship stock to you and having it shipped back. Shipping costs
back and forth would nuke the margins.
In General, mcheselka writes:
To help you decide if the ROI is acceptable, I've made a simple calculator:
I did it differently since all pounds are not created equally. One pound of
1x1 plates is ~1600 pieces and at 7 seconds that's just over 3 hours. (disclaimer,
I don't do used, only new set part outs so by the pound doesn't compute
in my brain.) I used 1x1 plates because I just wrapped up a 4 set (by 5 copy)
part out and when it came to the 1x1 plates I had 22 different colors in 22 different
tubs arranged on the floor around me like how your robot arm has them arranged
behind it.
What this sorter lacks in speed it makes up for in time, since it's designed
to operate unattended 24/7.
Totally spit balling here for things I would need and you should consider implementing.
Yes, the fact that it can run 24/7 is great but it would need constant intervention
in my situation. City Gardens for example has 1145 lots and if you tightly arranged
5x5 inch "cups" around the arm (assuming it's 20 inches long) you'd
have only ~39 individual cups.
A way I handle that is to use about 24 12qt tubs arranged around me as I mentioned,
then put all similar type of bricks in one and then sort those again later by
specific type and color. (think 1x1 and 1x2 modified plates.) They all go in
to one tub during the first round and then by specific part number and color
next.
A modification would be to have a matrix of "cups" on the top of the
table that lead to pvc piping under the table that directs the parts to their
specific tub. Then once the set is completely parted out, take those tubs and
resort them by tighter and tighter criteria.
It would be useful if I could tell it: I'm parting out 5x of 71741, pull
the inventory data from brickstore and then cross reference everything as it
goes. This would help with error detection if it knows going in that there should
be 10x 1x1 modified plates with horizontal clips (
but it put 9 or
11 pieces in to that bin then throw up a flag in the software that there might
be an over/under count in bin XXX. Ninety percent accuracy means you could have
just over 2800 incorrectly ID'ed parts if you did 5x of the 71741.
Another thing that may be helpful; before the program makes it's best educated
guess, have the sorting box bounce or jiggle once or twice to change the orientation
of the pieces. It looks like you had a
How does it handle picking up and dropping bricks/plates longer than 1x10? Off
centered with the drop point directly over the cup so that one end can hit the
edge and neatly fall in? What about those giant 8x16 plates?
In General, Ziegelmeister writes:
First off; I applaud anyone who's trying to perfect this for end users rather
then having us ship stock to you and having it shipped back. Shipping costs
back and forth would nuke the margins.
Yeah, the shipping is a killer. If it isn't affordable to get your own sorter,
maybe either a local physical store or a club can get one or two, and offer a
sorting service to people in their area.
I did it differently since all pounds are not created equally. One pound of
1x1 plates is ~1600 pieces and at 7 seconds that's just over 3 hours. (disclaimer,
I don't do used, only new set part outs so by the pound doesn't compute
in my brain.) I used 1x1 plates because I just wrapped up a 4 set (by 5 copy)
part out and when it came to the 1x1 plates I had 22 different colors in 22 different
tubs arranged on the floor around me like how your robot arm has them arranged
behind it.
Yes, I've calculated and use an average number of parts per weight. Like
you say, it's actually better just to think in terms of the number of parts
rather than volume or weight.
What this sorter lacks in speed it makes up for in time, since it's designed
to operate unattended 24/7.
Totally spit balling here for things I would need and you should consider implementing.
Yes, the fact that it can run 24/7 is great but it would need constant intervention
in my situation. City Gardens for example has 1145 lots and if you tightly arranged
5x5 inch "cups" around the arm (assuming it's 20 inches long) you'd
have only ~39 individual cups.
The layout of bins/cups will have a couple of default configurations, but it
will also be customizable (within physical limits) by the user.
A way I handle that is to use about 24 12qt tubs arranged around me as I mentioned,
then put all similar type of bricks in one and then sort those again later by
specific type and color. (think 1x1 and 1x2 modified plates.) They all go in
to one tub during the first round and then by specific part number and color
next.
The sorter will sort in any way that works best for you.
Ninety percent accuracy means you could have
just over 2800 incorrectly ID'ed parts if you did 5x of the 71741.
Yes, and what that ultimately means is the sorting process is a little slower
overall, since the sorter (machine or human) will have to re-sort the objects
that it incorrectly ID’ed. This is one of the reasons I say that this is a sorting
*assistant* and not a replacement.
It would be useful if I could tell it: I'm parting out 5x of 71741, pull
the inventory data from brickstore and then cross reference everything as it
goes. This would help with error detection if it knows going in that there should
be 10x 1x1 modified plates with horizontal clips (
but it put 9 or
11 pieces in to that bin then throw up a flag in the software that there might
be an over/under count in bin XXX.
Another thing that may be helpful; before the program makes it's best educated
guess, have the sorting box bounce or jiggle once or twice to change the orientation
of the pieces. It looks like you had a
I’m counting on feedback like this from users to help me develop future versions
of the software to give it these kinds of capabilities.
How does it handle picking up and dropping bricks/plates longer than 1x10? Off
centered with the drop point directly over the cup so that one end can hit the
edge and neatly fall in? What about those giant 8x16 plates?
There will probably always be some physical limitations (like object size), but
users will know this beforehand, and the system will let them know if they forget.
Some estimation about the price for such machine can`t hurt anyone right?
In General, mcheselka writes:
posting links to yt videos or other business-related info/sites is against the
rules of this public forum as far as I can tell -- and therefore out of courtesy
and to keep my account active, I won't. If you're interested in the
sorter please PM me and I'll be happy to answer your questions and send you
links to demo videos.
Cheers,
M
In General, Nubs_Select writes:
In General, mcheselka writes:
The automated sorting assistant has exceeded a 90% accuracy for identifying both
object types and part numbers. Sort by any combination of size, color, type,
or part number. Deployment is no longer a matter of “if”, but just a matter
of “when”. Sort for yourself and/or others. More details available if you PM
me.
Happy sorting,
Matt
Out of curiosity could you post some rough cost estimates? Also would it be good
for example if I parted out say 100x sets and dumped all of 1 bag in each time
would it be efficient at sorting that in a “fast” amount of time? Also like the
others mentioned if you post demo videos on yt or such then you may have interest
from people
posting links to yt videos or other business-related info/sites is against the
rules of this public forum as far as I can tell -- and therefore out of courtesy
and to keep my account active, I won't. If you're interested in the
sorter please PM me and I'll be happy to answer your questions and send you
links to demo videos.
You can post these types of links. If you're that worried about it email
Admin Russell for a blessing.
USA, Arizona
United Kingdom, England
.
Australia, Queensland
Canada, Ontario
France, Nouvelle-Aquitaine
Netherlands, Noord-Holland
