Challenge with part feeding

Etienne_Samson · October 2019

Hi Pros!

We have all experienced problem around part feeding to the robot, most of the time it's the blocker in our projects. Several solutions are available on the market, using jigs & fixtures, using bowl feeder, vibraiting table, using 2D vision, 3D vision etc.

I'd like to hear from you today is how do you select them, what do you use, when and why. Have you tried doing your own? Do you use standard solutions? How do you choose to go with vision or not? What are your biggest challenge with part feeding?

@matthewd92 @Enric @Ryan_Weaver I'll tag you guys since you are my heroes but I know a lot of people will have to share on this.

Merci

matthewd92 · October 2019

I guess the short answer is we’ve done about all of what you mentioned. The thing that we constantly strive for is to start as simple as we can, prove that the main task can be accomplished and then add complexity as we deem it necessary to increase unattended run time.

As we have experimented with different systems we have settled as our go to solution is tray type systems where we are putting parts in a grid on a tray and picking from this. These are generally systems that we can have laser cut from aluminum or steel plates so it’s cost effective and fast to get started with.

Ryan_Weaver · October 2019

Walking up to a machine and watching a person hand-load parts that are in a cardboard box lined with plastic is one of the first signs we're up against a difficult challenge. I think we try all of the things that most people would:
- Can the objects be easily stacked? Maybe on a rod?
- Could they be loaded into a "PEZ-dispenser" type tool?
- If they're rods, could they be singulated with an elevator tool, that pushes one up at a time?
- If they won't interleave and stick together, would they be a good fit for a vibratory table like Asyril?
- If none of the above, can they be loaded into a grid pallet?

Like @matthewd92 was saying, we always go to the simplest option first. The problem is that many times a customer doesn't see value in the robot if they first need to have someone take parts from a box and put them into a grid, or stack them up. Instead, they could just have that same person load the machine. So, we trend towards automated solutions, but they all tend to be custom in some way.

No, we don't typically design them ourselves. It's not really what we're suited for. Sometimes we'll do a custom tray for an Asyril feeder, but we almost never do pallets, stacking tools, or other custom designs. We occasionally leverage small engineer-for-hire or integrators for that kind of work, but it's not a smooth process.

Etienne_Samson · October 2019

@matthewd92 @Ryan_Weaver thanks guys, you've definitely seen a lot. What's the most common thing you consider a "real" challenge, unpacking stuff wraped in plastic (unpacking) or picking stuff from a bag full of parts (binpicking)?

Ryan_Weaver · October 2019

@Etienne_Samson - I'm not sure I've ever seen an application unwrapping parts, but the bin-picking scenario does come up. There are no tools on the market today that will reliably fit any part that we can just strap onto a robot to do bin picking. They are all ~50-90% functional.

It would be great if we had some kind of "Erector Set" kit (https://en.wikipedia.org/wiki/Erector_Set in case you didn't have these in QC growing up) that could build feeders for common part shapes or sizes. It's probably asking too much, but if we could feed 80% of the parts we came across in the field, we would undoubtedly sell more solutions.

Discussion

Challenge with part feeding

Comments