distinguish between plastic and glass

[recreators]

Hello everyone .
I am working on a project, and I stuck in one part, concerning, how to distinguish between plastic and glass, and papers if possible.
what kind of sensors can I use in this case.
thank you in advance

 

[dknguyen]

Is either the plastic or glass clear in the visible spectrum or infrared? If only one of them is clear and the other is opaqe in one of those spectrums it's not too difficult if you use LEDs and photosensors.

 

[gophert]

Hello everyone .
I am working on a project, and I stuck in one part, concerning, how to distinguish between plastic and glass, and papers if possible.
what kind of sensors can I use in this case.
thank you in advance

Glass, more dense than plastic

Glass more transparent than most plastics in UV

Glass, different refractive index than plastic

Define what type of glass and plastic.
Color of glass and plastic
Shapes?


Flotation is an easy way since glass is generally more dense. Floatation can be as wind sifting or literal floating - add a heavy salt to the water to increase density and make most plastics float and all glass sink. A 40% aqueous solution of CaCl2 is about 1.4g/ml (mist plastics are less and will float, glass pieces will sink (unless they are sealed with air to make them buoyant).

 

[MikeMl]

Dielectric properties are likely different

 

[recreators]

Hello , thanks for all answers
to clarify the idea, I want to make a smart trash that be able to sort these components. it's a small project to do at home

 

[gophert]

Hello , thanks for all answers
to clarify the idea, I want to make a smart trash that be able to sort these components. it's a small project to do at home

A local company has attempted this in many ways and ulimately had to resort to a three-stage classifying and sorting that included a vision system.

See https://www.cleanrobotics.com
Their device is called TrashBot

 

[large_ghostman]

I tried to upload a file but couldnt find it, if you look at bottling plants with glass bottles and plastic parts like caps, these all have systems on the production line to spot plastic/glass. I had a file from a system on a production line, extremely fast system but they used several methods, also they have the advantage of knowing the glass type and plastic type.

Optic is a good way, silicon labs have some good light sensors or use normal leds of different colours and use them as sensors, shine a bright white led through the material and see what values you get from the other leds. Some interesting experiments like that.

Also Mike has a good point on the dialectic value, but not sure where you would set the start and end point for each.

 

[dknguyen]

I didn't realize it was something as unpredictable as trash. I thought it was more like differentiating between a plastic, glass, or paper wall obstacle or sheet on a conveyor belt.

If it's a smart trash can where you can't control object shape or orientation, then you have to decide how you want to be able to toss stuff into it. (One at a time vs just a big heap of anything).

If it's one at a time, I think the easiest thing you could do is have a test chamber that weighs the thing and listens to the sound of the object as it was dropped in. Make the chamber of some material that makes a very different sound when glass hits it compared to plastic (probably glass or metal...some stuff material. Plastic hitting glass sounds too similar to plastic hitting plastic so don't use plastic). You can use a processor to do a FFT on a bunch of different test materials and sort of look at it and narrow down some characteristics to pick out in the sound.

You could still use optical I guess as a backup if you make a chamber that can predictable get the trash thrown into a predictable orientation...like a sloping V chamber with a sensor array so at least one will pick up the piece of trash. Optical is tough though since paper could be sheets or paper boxes. But plastic is much worse since it could be bottles, sheets, boxes, or crumpled up balls of plastic wrap. Bottles, whether glass or plastic also probably have labels which would probably rule out the use of optical.

If it's just a heap of anything, that's....much harder...

 

[large_ghostman]

Actually i missed something, its been all over the news here recently, not sure if its Norway or Sweden but the shops have a plastic sorting machine. You take your plastics in and shove it in the machine, it sorts the plastics into types and you get a deposit back. Its something they are thinking of doing in the UK, so maybe look into how they sort the plastics in these machines, if you can find a way to sort plastics then anything else isnt plastic (if you see what i mean), must be loads of patents etc that will give you a good idea on technique.

 

[MikeMl]

How about just blowing a stream of air diagonally across a moving conveyor belt? Glass will stay on the belt; paper and plastic will blow off the belt. If you have several air streams of progressively increasing velocity, you will like wind up with a gradation of densities, with dense-est stuff staying on the belt the longest.

 

[Pommie]

Wind separation seems to be the easiest solution. Run along a conveyor, fall at the end through an air stream, bins at the bottom for different densities. Ahhh, MikeMi beat me to it.

Mike.

 

[atferrari]

It seems that doing this would require someone teaching your system to distinguish shapes, colors (and dimensions?).

Good reading:
https://www.nbcnews.com/mach/scienc...g-one-dirtiest-most-dangerous-jobs-ncna866771

 

[tomizett]

You take your plastics in and shove it in the machine, it sorts the plastics into types and you get a deposit back.

I have a feeling that these "automated deposit return" systems just work by reading the product barcode on things like pop bottles that are put into them. I could be wrong though...

 

[gophert]

I have a feeling that these "automated deposit return" systems just work by reading the product barcode on things like pop bottles that are put into them. I could be wrong though...

In the US they do. You can only get a refund on a Michigan coke bottle if it has the Michigan coke bottle bar code. Same for NY and other states. Otherwise you end up with a Kramer from Sienfeld filling a stolen postal truck with cans from NY and driving them to Michigan for the bigger refund.

 

[large_ghostman]

I honestly dont know, but seeing as its likely to come here, i will try and find out as the tech itself is interesting.

 

[BobW]

In the US they do. You can only get a refund on a Michigan coke bottle if it has the Michigan coke bottle bar code. Same for NY and other states. Otherwise you end up with a Kramer from Sienfeld filling a stolen postal truck with cans from NY and driving them to Michigan for the bigger refund.

That's the rule here in Canada too, but they don't check the barcode where I live; the time spent doing that would cost them more than they'd save. They just have a rule of $50 max total refund per person per week (or month? I can't remember). That makes it too unprofitable to import bottles and cans.

One thing about sorting different trash materials is that it will be much easier if everything is similar in size. If it's just bottles, then that should be fine. But if you're talking about any possible size (eg., living room picture window), then you may have to consider some method of breaking things into more practical sized pieces.

 

[large_ghostman]

I was under the impression barcodes were not involved, but i didnt actually pay much attention to the news article at the time, hence why i cant even remember the country they gave the example for! Its one of the Scandinavian ones however, bit disappointing if it turns out to simply be barcodes, but i have a vague memory of clear no label milk bottles being shown, or i could simply be imagining that bit.

I have mentioned all this to my mum, who apparently remembers being a little girl and having two fizzy drinks lorries go around the town selling fizzing drinks from the vans, one was brand A and the other a different brand. They used glass bottles and would take only there own, from her memory they gave a pretty decent return on the bottles but they had to have the lids as well as the bottles. Some people would chuck these bottles out, and my mum got her pocket money by going around and checking hedges etc for the bottles!

I assume this was early to mid 1970's, she can only remember the name of one company Corona. So its all a bit closed circle now...

 

[gophert]

the time spent doing that would cost them more than they'd save.

In Michigan, most stores had machines that required the consumer to stuff cans, glass bottles and 2-liter bottles into and would generate a ticket with the value - which could be submitted to a cashier for cash or to offset the grocery bill.

 

[large_ghostman]

Anyone in the UK just seen the recent edition of the BBC tech news called 'click'?

I ask because it showed something that shocked me, we are talking plastic and glass in the thread and how to distinguish. They showed a retail center system using AI to map peoples behavior and shopping styles, part of it was the cameras actually looked at people and guessed age range, race and demographic type almost instantly! It then followed them around the shopping center. It put little labels above people, if you get a chance look out for it. If i find it on iplayer i will post a link.

Scary but fascinating, anyway the point is if they can do what the program showed then a camera system could easily be used with common libs to sort material out.

 

[gophert]

Anyone in the UK just seen the recent edition of the BBC tech news called 'click'?

I ask because it showed something that shocked me, we are talking plastic and glass in the thread and how to distinguish. They showed a retail center system using AI to map peoples behavior and shopping styles, part of it was the cameras actually looked at people and guessed age range, race and demographic type almost instantly! It then followed them around the shopping center. It put little labels above people, if you get a chance look out for it. If i find it on iplayer i will post a link.

Scary but fascinating, anyway the point is if they can do what the program showed then a camera system could easily be used with common libs to sort material out.

As. Said above, CleanRobotics.com does just that, image processing, and separating conveyors to identify cans (whole, crouched, ...) from plastic bottles (whole, partially full, crushed) from cups, or even a half of a hamburger. Each is sorted into organic waste, aluminum, steel, or plastics bins. All happens in fraction of a second.

It may not be a great (optimal) use of energy and recycling (their trash bins are actually in stadiums, convention centers and malls), but they have made plenty of cash by translating their vision system software into other sorting automation projects.