[Beginner here with no experience] Need help with Ideas for implementing an automation process

[pyronic]

Hi Everyone,


I need help building an automated tool that takes roughly 4000 fasteners belonging to roughly 100 categories (types of screws, bolts, washers, nuts) and categorizes and inspects every fastener and packages together and labels those packages as "Category A - working", "Category B - Defective", "Category C - Working", etc.
I need help with the mechanical & electrical design of this system.
You can assume categorization and inspection is done using some vision based modules.


I am sorry if my question is too vague but I am looking for directions on how one would go about building such a system.

EDIT: Adding some additional info that we have narrowed down after initial reseach:
1. Plan to use a bin picking robotic arm. (space constraint of 5ft by 5ft so can explore other mechanisms as well if they fit)
2. Total 4000 in every iteration. Idea is to have a recycling mechanism from old fasteners to reuse the ones in working condition.
3. We can expect one such iteration to be run once a week.
4. Expected inspection + categorization time ~60 secs per item.
5. We can't use magnet based methods since not all fasteners would be magnetic.
6. Idea is to build a Proof of concept with a budget of around 15000 USD (what kind of tradeoffs might this lead to? can it be done with a lower budget or would it need to be increased? )

Thanks a lot!

Regards,
pyronic

[tician]

Lots of options.

Easiest that comes to mind is a hopper holding the unsorted parts that are fed onto a conveyor belt in a well spaced manner (one fastener on the belt every couple inches). The first section of conveyor is solely to perform the visual inspection before heading into the sorting section. The sorting section is simply a bunch of bins set immediately beside the belt with paddles/diverters that push the part off the belt into the desired bin. The rest is just timing paddles/diverters to conveyor belt speed to get the right part in the right bin.

This is all assuming the visual inspection system is actually able to determine quality. If a bolt has been torqued down even once, then it has been stretched/damaged in a way that is unlikely to be detectable by any visual inspection.

[jwatte]

If all you have is 4,000 fasteners (not 4,000 fastener TYPES) then you should do this manually. It'll take two hours, and then you're done.

If you have 4,000 fastener TYPES that need to be sorted and arranged, and you have literally millions of actual fasteners, then you could perhaps build this. The hardest part is the vision recognition model -- training it on enough sample data of all the different kinds of fasteners is quite an undertaking. (90% of the work will be in creating, gathering, preparing, and labeling all the input sample data)

Then again, maybe that's the part that you have already solved? If so, I agree with Tician. Some kind of hopper and itemizing actuator that feeds out one at a time. Those are kind-of hard to make robust, though; you're going to needs some kind of mechanical "stirring" to keep them from wedging.
Also, if the different fasteners have significantly different size, you might want a sifting stage that moves the objects into approximate size categories first. This will greatly simplify the construction of each of the hoppers. You could have a sift with successively smaller nets in the center, and then lay out a number of processing lines stretching out in a star shape, here.

Are the fasteners magnetic? You may be able to selectively lift them off a conveyor belt using an electromagnet, perhaps on an arm or gantry that can move them to the appropriate bin. One gantry+arm+magnet may be able to serve all the bins for a single processing line, and let you stack bins in two dimensions, instead of putting a hundred separate paddles on there, each with a bin along the belt. It would probably be more compact.

[pyronic]

Thanks tician,

Based on our initial research, we are thinking of a bin picking robot due to space contraints. (Can explore hopper, conveyor, diverter mechanism if it can be done within a 5ft by 5ft space.)

Additional info:
1. Plan to use a bin picking robotic arm. (space constraint of 5ft by 5ft so can explore other mechanisms as well if they fit)
2. Total 4000 in every iteration. Idea is to have a recycling mechanism from old fasteners to reuse the ones in working condition.
3. We can expect one such iteration to be run once a week.
4. Expected inspection + categorization time ~60 secs per item.
5. We can't use magnet based methods since not all fasteners would be magnetic.
6. Idea is to build a Proof of concept with a budget of around 15000 USD (what kind of tradeoffs might this lead to? can it be done with a lower budget or would it need to be increased? )




Some questions on my mind are:
1) What kind of robotic arm I should go for? (5 or 6 DoF)
2) What are the available options to purchase this robotic arm? (Feature and cost comparison)
3) Basics of designing the other modules that can be implemented by a beginner
4) For bin picking can Reinforcement learning be used?

[pyronic]

It takes a lot longer than that due to subtle differences in fastener type.
At around 1 minute per item we want to implement a solution that does inspection & categorization in around 4000/60 = 66 hours.

Adding some additional info that we have narrowed down after initial reseach:
1. Plan to use a bin picking robotic arm. (space constraint of 5ft by 5ft so can explore other mechanisms as well if they fit)
2. Total 4000 in every iteration. Idea is to have a recycling mechanism from old fasteners to reuse the ones in working condition.
3. We can expect one such iteration to be run once a week.
4. Expected inspection + categorization time ~60 secs per item.
5. We can't use magnet based methods since not all fasteners would be magnetic.
6. Idea is to build a Proof of concept with a budget of around 15000 USD (what kind of tradeoffs might this lead to? can it be done with a lower budget or would it need to be increased? )

[tician]

Are these fasteners being used in any sort of life/limb threatening situation? DO NOT REUSE THEM! EVER!

The highest strength alloy steel bolts can be reused in some situations where they have not been torqued down anywhere near their limit, but softer metals like brass and stainless steel (those non-magnetic materials) are much weaker and easier to damage without any visible signs of damage. Visual inspection will only tell you if the fastener is rusted or already broken/severely deformed. It will not tell you if the faster has internal damage leaving it ready to break the next time you try to torque it down a bit.

We have to use stainless hardware at my workplace because the equipment gets exposed to ozone, humidity, and assorted spilled chemicals that make alloy steel rust in place. None of the equipment involves life/limb, but we cannot reuse most fasteners because all too often they break when installing and we then have to spend way too much time drilling them out and trying to repair the tapped hole.

If these 'recycled' fasteners are being resold (even with labeling as 'used'), then you are destined for lawsuits and/or prison.

[pyronic]

Are these fasteners being used in any sort of life/limb threatening situation? DO NOT REUSE THEM! EVER!

The highest strength alloy steel bolts can be reused in some situations where they have not been torqued down anywhere near their limit, but softer metals like brass and stainless steel (those non-magnetic materials) are much weaker and easier to damage without any visible signs of damage. Visual inspection will only tell you if the fastener is rusted or already broken/severely deformed. It will not tell you if the faster has internal damage leaving it ready to break the next time you try to torque it down a bit.

We have to use stainless hardware at my workplace because the equipment gets exposed to ozone, humidity, and assorted spilled chemicals that make alloy steel rust in place. None of the equipment involves life/limb, but we cannot reuse most fasteners because all too often they break when installing and we then have to spend way too much time drilling them out and trying to repair the tapped hole.

If these 'recycled' fasteners are being resold (even with labeling as 'used'), then you are destined for lawsuits and/or prison.

Hi Tician, You have an absolutely valid concern.
Here's a paper that shows a sample use case of a robotic system we want to buld.
What are the ways to identify bolts that have been torqued down?
Are there other check that we might miss through visual inspection?

[tician]

I'm inclined to suggest a guillotine for the penny-pinching executives whose bonuses cost more than flat out replacing all the fasteners during every aircraft airframe/engine overhaul. Barely costs a couple hundred dollars to build; maybe splurge on a wood chipper if you're in a hurry. Fasteners are mass produced, so there is always a risk of sub-standard items in a batch. Re-using fasteners increases the risk of sub-standard fasteners being utilized at too many critical locations. What does failure cost? When fasteners fail in our equipment, there are no human costs beside the time wasted performing repairs and even that is more than the cost of a few dollars of new fasteners.

When a bolt is torqued down, it stretches. That stretching is very small and reversible, until its not. Because of manufacturing tolerances, there is no way to visually identify a bolt that has been pushed into plastic deformation unless it has already begun severe failure (very severe necking between threads and/or fractures). You would basically have to repeatedly and precisely measure the length of every single bolt at a known temperature and keep that in your records, then measure again when it is removed during overhaul. And even with all that, you will likely have repeatability issues because of the very small changes in length involved. Even if there is no significant plastic deformation, the fastener may undergo strain hardening making it more likely to fail after being tightened down again to the same torque.

Ultrasonic imaging or high-energy (x-ray, etc.) imaging may be able to detect fractures forming inside the fastener, but might not be able to identify plastic deformation unless the grain structure has been severely altered.

[tician]

Just a little clarification: a fastener is tightened to a specified torque to exert a desired clamping force. There are a number of parameters that must be examined to correctly select a fastener (size, tensile strength, thread depth, coatings, torque, etc.) to safely meet the clamping force requirements. Improper design or improper 'equivalent' replacements can lead to terrible results.

When a fastener leaves the elastic region and enters plastic deformation, then it has by definition failed. That small amount of plastic stretching means it will not exert the desired clamping force again until it is further tightened, but just tightening it again risks fracture ('silent' failure becomes catastrophic failure).

[jwatte]

There are cases when I re-use fasteners.

For example, when I assemble a 80/20 based trade show booth frame, I re-use the bolts and studs that go into the frame.

When I re-hang a picture on my living room wall, I re-use the bolts. I also do this for the TV screen mount, and for the speaker mounts.

What do all these use cases have in common?

1) The fasteners are generally ludicrously over-specified for the job they do
2) The cost of failure is, at worst, buying a new $500 TV, or having new trade show booth background inserts printed

If the re-use case is similar to these, then it's totally a fine thing to do.
If it's anything to do with repeated maintenance or refurbishing on heavy machinery or vehicles, then I'm with Tician: safety is more important than pennies.