RFID or alternative for short-distance setup

Apologies for the delay on all this, was in Cyclone region and got very distracted from this project. But now ready to go!

Thanks for all responsess.

Let's presume the scoop is associated with a person already, to elimiated that part of the complexity. Say if they're numbered (no tech, just written on each scoop).

Does that mean there's a limit of just 80 unique ID's using those UIC stickers? I think 5cm isn "just" workable (but absolutely no less, ideally it's 10cm)

What would "read" the sticket on the scoop?

I'm not sure on your idea of a Pi/Camera at each destination. That seems like a complex setup that would require power too at each bucket no?

Agaion, load cells not an option as you mention- too much expense and setup complexity

 

juzz's design has powered readers on each bucket.

If you are putting RFID scanners on the buckets, you can have an unlimited number of tags on the scoops and you won't need a programmer. Just buy the cheap, non-programmable tags and they will come with a unique ID that you can associate with the scoop in the backend.

What sort of reliability are you expecting? Is one missed scoop per user per day going to cause a problem? 2 missed scoops? 50? Is counting 2 scoops when it was actually 1 a problem? Is it a problem if the user can deliberately increase or decrease their scoop count?

A lot of the ideas suggested here will work to some extent, but if you are stuck with the "scoop goes into a bucket" approach (as opposed to using some sort of dispenser mechanism) I don't think it's possible to create a solution that doesn't have some weak spots so the details will depend heavily on what sort of compromises you are willing to make.

 

^ Solid post. Great questions.

It'll save you some effort if each RFID belongs to a person, and a scoop. And the same person always uses the same scoops.

Example:
You employ Joe, John and Jane to do the job.
Joe uses red scoop 001, blue scoop 002, and yellow scoop 003 - every time he works.
John - red 004, blue 005, yellow 006 - every time.
Jane, red 007, blue 008, yellow 009 - every time.
When you read the scoop at the bucket, you'll know who it was, and what colour it was.

I still don't think you'll dodge needing power/a reader at every bucket. Batteries might work for the power bit for a shift, but you're going to need to know what went where - so each 'where' (bucket) needs to read a 'what' (scoop).

 

The original question specified powered scoops and unpowered buckets, which is definitely possible but with different trade offs.

 

If you don't care about weight, just how many times a scoop of whatever has been removed, why not just put a latched lid and a momentary switch on each bucket and just track how many times it's opened? You'd have to work out a reset system for when you refill but probably easier than some IR\RF solution. You could assign a bucket to each user rather than using the scoop registered to a user or NFC tag the bucket lid.

 

Yes, having powered at each bucket really isn't an option. They can move about, are subject to being washed etc. So ideally the buckets stay as simple/low tech as possible. This also means the

I like the idea, but as per the above (keping buckets simple) it won't work.

I think we can have a ~10-20% inaccurracy on readers. Does this make the task any easier?

Understood, but keen to find a work around.

Where lies the problem of having the readers on each scoop and keeping buckets passive? Is it about the distance of the tag to the scoop (eg 10cm)?

 

Nah, you're right - you can pick which end the power is needed at easily enough - it could be input (scoop) or output (bucket).

My logic behind picking the output is that it is likely to be the most stationary, so easier to power and network reliably (assuming you don't want staff carrying battery packs and your shed is electrically noisy, which will impact the cost of a WiFi deployment).
Keep in mind you don't necessarily need to power the bucket itself - just the position that bucket usually resides in.

Whatever you ask to do the 'reading' also needs connection to the database - so you're likely to need that robust WiFi solution through all the trodden paths in the shed if you're asking the scoops to report home, rather than just making sure your receiving positions have coverage if you go the other way.

As heydonms says above - it's possible either way, you'll just have to work around the trade-offs. I see less trade-off going with powered receivers, but it's difficult to visualise without a process map

 

The number one problem that I see is a lack of consistency. The size of the opening on the bucket is potentially smaller than the distance between buckets, or the depth of a bucket. It is extremely easy to reliably detect when a scoop enters and leaves a bucket, but I believe it will be practically impossible to do it without the user's cooperation, while also avoiding false positives from buckets that are near to the target bucket, or false negatives/false retriggers when the bucket is nearly empty.

You could obviously improve consistency with dispensers, they have to put the scoop in the right place to get a read before anything comes out.

IMHO, the easiest fix is to drop the "no user training/cooperation" requirement. You must already be training the staff to a certain extent to make sure they aren't feeding chicken supplements to the cows, planting fertilizer, or swapping the scoops between users, so add "Touch the handle of your scoop here and wait for the beep before scooping" to your training and everything becomes a million times easier.

I still think ibuttons would be a great solution, they are waterproof, self aligning, and the logging hardware is available off the shelf in the form of security patrol logging devices, but once you have people deliberately presenting the device to the tag, or vice versa, 90% of the complexity goes away regardless of whether you are using ibuttons, NFC, LF RFID, barcodes, etc..

 

All valuable inputs for thought, thank you both

Could I get around this if the "scoop" keeps the readings, eliminating the need for wifi? Then when the scoop is returned "home" it has a list of what "scoopings" have taken place?

What about using NFC/RFID blocking tape like this? So it keeps all the signal underneath?

I see your point on the simplicity there, but sadly it involves (even if tiny) training which isn't possible (as often it will be strangers). Maybe the help could be a some kind of smart latch that unlocks only if an iButton is inserted? Mechanical of course (no moving parts....)

 

I think you're thinking about RFID like it's some kind of liquid and if you build a container for it you can neatly define where it is and where it isn't, and then you can get reliable data on whether the reader is "in" the RFID area or not. That's not how it works. RFID uses a field that is absorbed, reflected, and detuned by metal in ways that are impossible to predict without doing complex modelling. If you stick an RFID tag on the side of a container and then wrap it in foil, you might get good reads inside the container, or you might detune the tag so you get no reads, or you might end up with nodes where it reads perfectly and other areas where there is nothing and the location of those nodes might vary depending on the angle that you are holding the reader, and your scoop could pass through several of these nodes (or none of them) while you are moving it around.

I still believe that what you have described isn't possible. There is no solution where you don't change your delivery mechanism, or provide training, and you get something approaching a reliable count.

You just let strangers wander on to your property, with no discussion of what they need to do, how they will get paid, where the toilets are? You just silently hand them a scoop and point towards the shed? If you're going to need to change something about your process in order to achieve your goals, changing this still seems like the simplest option.

A mechanical latch with no moving parts, and no electricity, that is unlocked using a digital signal? It seems unlikely.