Public platic bottle/glass bottle/can accept machine.

Public platic bottle/glass bottle/can accept machine.

Hello i'm a student and trying to build some kind of machine to accept bottles and cans for recycling for my project.

I need to sort the bottles in different bags depends on bottle type (bag for metal cans, bag for small glass bottles, bag for large glass bottles like wine bottle, bag for plastic).

My first problem is to detect the bottle type when inserted in the machine, I guess I can detect the cans with metal detector, but how to act with the glass or plastic...

After the bottle type is detected I need to put it in the correct bag (with motor and some kind of trail to take the bottle to the correct bag).

If someone puts something else then a bottle or a can (lets say someone inserts a wood or rock to the machine) I need to throw it back.

I need to count the bottles (each bag can contain some exact number of bottles or cans, so if the machine accepted already 200 cans for example, its need to stop accepting cans until the bag is changed and return it back to the person who puts it in the machine).

In the end I need to print some kind of thermal reciept to each person that put the bottles, he will get money with that reciept.

So I need to count the total amount of bottles and to count the returned bottles for each person seperatly to print the reciept.

I need to be able to reset the count of total cans or bottles after the bag is changed by the operator (when the bag is full with the 200 cans or any other amount) depend on the bag type.

I thought to use arduino to operate all that with motors to move the bottles, do outside lcd and button to let people know that each bottle is counted and let them press the button when they finished to insert all their bottles to get the reciept.
I need to add some sound bazzer when the machine is full or wrong material detected.

I want the machine to be wifi or ethernet connected to report the operat when bag becomes full and stuff like that.

I thought to use scale to take bottles weihgt to detect the bottle type but sometimes its half full of water and sometimes broken so its not an option to detection.

There is machines like that in the market but they detects the bottle type by the bottle ean-13 barcode, and if the barcode is missing or unknown they not accepts the bottle at all.

And I need to write down the current numbers in the machine so if it gets power off and on it will remeber what was the last numbers.

Its a big project so for now to detect that there is a bottle and to identify that its a bottle and not a piece of wood/rock its a start.

Big thanks for everyone that can help me with something!!

This is a HUGE, multi-year, expensive project. You'll have to build multiple prototypes (and parts of it). The mechanics and detection are going to be the hard part. So without some decent funding and a team of engineers bringing in the various skills - well, not to discourage you, but you better forget about it.

Metal: easy, metal detector. But do you want to distinguish between iron (from canned food) and aluminium (beer, soda cans) as well? Iron is magnetic, aluminium not.

Glass/plastic: maybe IR works, but more reliable would be x-ray, and it should have less of an issue with labels being in the way. That's at least a way to distinguish plastics (HDPE, PP and PET are most commonly used for plastic bottles - you want to accept and sort out all those? How about the colours of the bottles?). As a human I'll be easily able to tell them apart by looking at them: milk jars are HDPE, shampoo bottles PP, water and soda bottles PET. But they also may be PVC, and you REALLY want to keep those out or the value of your PET fraction plummets.

Then the mechanics of it all. Very hard to make it reliable. Glass bottles may be broken, many different shapes to deal with, bits of shampoo and beer running out of them making a mess of the internals, etc.

All the other things you're asking for are easy. Counting, storing counts over a power outage, bar code reading/identification, receipt printing, etc. Even the control of the various motors based on your sensor input should be quite straightforward to implement.

Thanks for the answer, I guess I will start with the aluminium cans, they pretty much same sized so I can limit the machine bottle entrence hole to that size, and I can verify them with the metal detector.

Should I use servo to block the bottle entrence after 20 mm to let the metal detector to detect and open close it for each can?

How can I verify that the can did fell down in the pipeline and not just been hold in the entrence and to be sure that I can do count + 1.? Should I use some ultrasonic sensor to know that the can entered after the servo door opens?


Optical may work better for that.

You also have to make sure that the gate is not closed when there's a hand in the opening or so, even if the can has fallen in the bag.

How can i verify that there is no hand inside? Some infra red sensor to check that the entrense is free?

Well, that's part of your challenge.

You indeed need some kind of sensor that tells you the entrance is free. Maybe also a sensor on the lid that closes so it detects there's something in the way and opens again. There are probably more safety issues with a machine like that, which you have to take into account. It offers money for deposits, so you bet people will try to insert all kinds of items, if only to try it out.

Oh and another thing: if one of the bags is full, what are you going to do? E.g. the glass container is full and the user puts in another glass item. Return it to the user? How are you going to do this return? You can't just spit it out (it falls and breaks). You'll have to find a way to present it, and then detect the item has been taken back.

What is your budget?


About the budget, I guess 1-2k for the electronics, I'm checking to estimate what needed, Im open minded about it.

About the full bag, the plan is to do the gate 15-30 cm deep so it will be possible to just take the bottle back, if the bag full and someone inserted something its just wont open the gate and will sound a bazzer untill it tooked out... thats why I need to identify the object beffore the gate is oppened... if the gate is opened and he put something that blocks the safety sensor and keep droping bottles it will stuck the machine and wont be possible to give it back to him...

Maybe I need two gates..? So the bottom gate wont be reachable by hands and can be closed free while the entrence gate will be closed only if the safety sensor is free... so if the safety sensor is busy and bottle droped it will stuck in the bottom gate... ?

Its a little problematic....

$2,000 would be a very small fraction of what would be necessary.

You need drivers, motors, servos, conveyers, welding, sensors, equipment and personel protection circuitry and mechanisms, material detection circuitry, deflection equipment, numerous power supplies,wiring harnesses and other hardware.

Not going to happen on low budget with limited time input.

1-2k - that's USD?

Add a zero or two for a more realistic budget. Plus a year of your own time.

If all you need to detect is metal or non metal, then you need two sensors.

One sensor senses all samples, metal and non metal.
Second sensor senses metals.

If your sample trips both sensors its metal.
If your sample only trips the First sensor then it non-metal.

A simple door in the floor of the chute after the sensors.
It opens if a metal sample appears and it falls through to a metals chute.
It stays closed if its non-metal and goes off the end of the chute.

Shouldn't be too hard to do a proof of concept in the lab.

If you need to identify steel over aluminium, then another detector under the metals chute to do the same thing.

Tom... :slight_smile:

To stop arms, hands etc, consider an L-plate that flops as each item enters the chute - keep person away from any moving parts!.

A geared servo to steer the output chute director.

Possibly a second ‘drop plate’ to release the item when the output is in place.
Check for obstruction, or flaps being obstructed - fail or ‘dump’.

Sensing.. tricky but not impossible.
Weight, optical IR opacity, colour, magnetic characteristics - need to think a bit more.
All the logic, memory and control are fairly straightforward - just need testing.

Biggest issue will be mechanical reliability.

Thanks for your answers! I will consider it all :slight_smile:
The biggest problem is the identification.. its hard to identify that its a bottle and not a glass of watter or something..

Indeed its easier to detect the cans.