We do all of our picking in reality by ship to, so pick slips would be nice if they were grouped by pick slip, then by part, then order line and release. The issue we run into is trying to find all parts from multiple orders across the vanilla pick slips so we don’t keep going back to the same bin multiple times. If we went order by order and order 1 has part A and order 5 has part A they would like to pull both those order lines for the same part at the same time. I’m not sure it’s the best explanation but basically batch picking parts for the same ship to.
I see what you are saying.
I am sure we are also doing that, haven’t ever thought of a solution though.
I’m thinking either rebuilding pick list report grouped by CustNum ShipToNum and working down from there, or building a dashboard that they can use for picking that is grouped that way.
That is where my mind went. So long as it support the process of picking.
Is the team out there doing the picking okay with changing the process by which they pick?
Does this new way create any new, difficult scenarios that they don’t have to deal with right now?
I see how your solution would make it more efficient, but whenever new problems are created I typically see people get hung up on them even though the process is more efficient.
I like your idea of rebuilding the report and/or using a dashboard. It sounds promising. Let me know how it works out.
No they never are
Couldn’t tell you i’m just trying to solve the issue of them printing 10 sales orders off and miss-picking because it’s hard to keep track of what’s been picked for what order from what bin. Over/under shipments are happening too much. Just in the planning phase trying to come up with a solution. We have already rebuilt packout from the ground up we’ll get there.
I have seen you build some awesome tools for your company.
No doubt you are going to get there with this solution too.
I am facing the same issue here where we will ship the wrong thing or too little or too much…
In my mind I want to train an AI/ML model to know, based on camera feed, that the operator picked x Qty of Material A for Order 12345 and then keep track of that.
For me it was just skids Joshua, not necessarily the materials, that I was trying to keep track of. Sometimes they forget to put a skid on a truck or they put one too many on there.
Pretty much the process would be:
- Start a pick, tell system what order you are picking (it would then go out and retrieve aggregate list of everything (packs in my case) you need to pick).
- Use live stream with OCR recognition to tell when something has been picked and how much.
- Have user say they are done with the picking.
- Program will look back through the recorded packs based on OCR and tell user if too little or too much has been packed.
Barcoding could do this well too where you scan everything you pick, but I would like OCR to save the operator from having to do the scan action.
I trained an Azure OCR model to be able to identify a packslip of ours and record the packnum and order number as data fields. I was then messing with API calls to that service to try and reconcile the pick list with what has been loaded to be able to tell if anything was left off.
That’s about as far as I got, but I swear it is possible. Pretty much my take away from insights this year was, AI/ML can augment a human’s ability to do a job. I always thought automation, but it truly can help them pick more accurately too.
I think barcoding what you are picking as you are picking it would ensure a proper pack as well.
Imagine if you had to scan a barcode every time you picked something… There’s no way you could get it wrong unless you didn’t put it in the box or on the skid…
It would probably slow down the process a little, but it would ensure accuracy.
When they are picking in the qtys we’re talking, if this were the case, I’d have a revolution lol
lol I know man, it’s just not practical.
That’s why I want to use AI/ML video recognition to do it for them.
It can analyze video and understand what is happening if we train it to do that.
I was close to coming up with a proof of concept and if I ever get there (trying to study this over the winter) I will for sure get it to you or show it.
I struggle with the application development side of it. Using the Microsoft azure tools was pretty fun and simple once you learned where and what all the settings did plus their API documentation is nice.
We tie our Material request reports by order. When I customized it, it will print out with a barcode so we can use Docstar to pull those back in and tie it to the order complete with any notes from the pickers.
We still pack out by ship to using PCIDs, but since we receive all day long, we can’t really keep from multiple trips to the bins…
2 Likes
Cool solution Doug!