ROBOFORMING: The Future of Metalworking?

[David_M]

What do you think?

 

[Rob F.]

Wow, that is a pretty cool way to work sheet metal.

 

[boosted]

I first saw this a process a little over a decade ago. I believe it was a Ford/Boeing joint venture at the time.

I think the capabilities of the process are too limited to be revolutionary, and IIRC the cycle time was so slow that for most prototype applications (in aerospace) it made more sense to 3D print a disposable die, and then hydroform the part.

Sheet metal continues to be a real PITA, and there is a pretty hefty market if somebody develops a way to go direct from CAD to part for one off parts.

AFAIK the fantasy of 3D printing a fender is still pretty far out.

 

[jaguar36]

The cycle times don't sound too bad for big parts in low quantities. I don't understand what you would do with them though with that awful surface finish. Certainly not any sort of aerodynamic surface. Can't image anyone would want them for any sort of aesthetic purpose either, unless they could work the ridges into the design.

 

[empower]

i wouldnt call it THE future. but part of the future? certainly

 

[empower]

The cycle times don't sound too bad for big parts in low quantities. I don't understand what you would do with them though with that awful surface finish. Certainly not any sort of aerodynamic surface. Can't image anyone would want them for any sort of aesthetic purpose either, unless they could work the ridges into the design.

but golf balls are aerodynamic with dimples! :P

 

[standardparts]

The cycle times don't sound too bad for big parts in low quantities. I don't understand what you would do with them though with that awful surface finish. Certainly not any sort of aerodynamic surface. Can't image anyone would want them for any sort of aesthetic purpose either, unless they could work the ridges into the design.

"incremental sheet forming".....search it on youtube. Been used for about the last 10 years for prototyping with differing results. Remember Ford showing it off about ten years ago.
Took a look and now seem it's come far enough along that it's use is becoming more accepted.
"incremental sheet forming" pretty interesting and looks like it's capabable of producing nicely finished parts.

 

[Orange Vise]

The cycle times don't sound too bad for big parts in low quantities. I don't understand what you would do with them though with that awful surface finish. Certainly not any sort of aerodynamic surface. Can't image anyone would want them for any sort of aesthetic purpose either, unless they could work the ridges into the design.

Bondo to the rescue!

 

[MCritchley]

I hope the outfit in the video succeeds as this will fill a big gap in the prototyping world. Most companies will not invest in building a sample die due to the costs involved, this technology should be much more affordable for prototyping efforts.

 

[gustafson]

I was kind of 'whatever' until the guy said they could form parts faster than they could simulate them on the computer

OK, that might be useful

 

[jaguar36]

I was kind of 'whatever' until the guy said they could form parts faster than they could simulate them on the computer

OK, that might be useful

Pretty sure he meant simulate forming it, which isn't suprising, that is a very complicated process to simulate. Simulate the part in use, no way. Sheet metal parts are super easy to model using FEA.

 

[Strostkovy]

I first saw this a process a little over a decade ago. I believe it was a Ford/Boeing joint venture at the time.

I think the capabilities of the process are too limited to be revolutionary, and IIRC the cycle time was so slow that for most prototype applications (in aerospace) it made more sense to 3D print a disposable die, and then hydroform the part.

Sheet metal continues to be a real PITA, and there is a pretty hefty market if somebody develops a way to go direct from CAD to part for one off parts.

AFAIK the fantasy of 3D printing a fender is still pretty far out.

I 3d print dies and stamp 1/16" aluminum. It works great. I'm setting up a router table to machine dies from block faster.

 

[GiroDyno]

Could be useful for pre-roughing in the shape of molds for aerospace tools out of materials that are expensive and/or hard to fabricate and machine. We looked into this about 10 years ago as a substitute for welding up and then machining down slabs of Invar to make 787 fuselage tools. Was cost prohibitive then, cool to see it again, and interested to where it goes from here.

 

[Ox]

As for the surface finish, I agree with y'all, but I am sure that if your app required a better finish, that they could Shirley shorten their step-overs.

But if you are looking for fitment and whatnot, then maybe a rough but true surface would be quite sufficient.
Some of the cheap 3D parts that I've seen didn't look any better'n this.


--------------------

Think Snow Eh!
Ox

 

[garyhlucas]

Metal spinning, not round. Don't iunderstand using a robot. Seems like a ridgid frame holding the sheet and say an X axis on each side of the frame with a very stiff Y axis running left and right carrying a two Axis with the tool would make more sense. Much less complex kinematics and much less deflection.

 

[Ox]

I would fully agree with that.
A simple 3x gantry on both sides would likely be MUCH better....

... BUT!
The salesman has a MUCH easier time "OH WOW'ing" the customers with this set-up for sure!

The only real catch to the gantry would be not being able to hit the part of the roller that you might want to.
Now, I guess the real question might be - how important is that really?
You're likely never going to be working at dead straight - like you could be with a ball and mill, BUT being slightly off center could end up with the ball skidding?

Or maybe a bit different type ball end effector?
Sumpthing along the lines of the ball drive off of a 1980 video game? (Missile Command?)
Upside down 2000 era mouse.

Would seem to require much less computing power, and would likely round those corners much faster?


-------------------

Think Snow Eh!
Ox

 

[Strostkovy]

I would fully agree with that.
A simple 3x gantry on both sides would likely be MUCH better....

... BUT!
The salesman has a MUCH easier time "OH WOW'ing" the customers with this set-up for sure!

The only real catch to the gantry would be not being able to hit the part of the roller that you might want to.
Now, I guess the real question might be - how important is that really?
You're likely never going to be working at dead straight - like you could be with a ball and mill, BUT being slightly off center could end up with the ball skidding?

Or maybe a bit different type ball end effector?
Sumpthing along the lines of the ball drive off of a 1980 video game? (Missile Command?)
Upside down 2000 era mouse.

Would seem to require much less computing power, and would likely round those corners much faster?


-------------------

Think Snow Eh!
Ox

Inverse kinematics is complex but really only for determining the formulas. Once the motion equations are figured out it's pretty easy for a computer to do the calculations. I suspect long simulation times are just due to poor software optimization. Which isn't an insult; software is costly to optimize.

If I were to try this at home, and maybe I will, I would just use dual XYZ systems poking at each other. Then you can offset the balls for the required angle at the contact point. I'd be somewhat inclined to try an elliptical roller with a rotary axis, but if not something like a ball transfer or at a minimum a lubricated tool would be my next try.

If only I had a shortage of projects...

 

[jaguar36]

I think you need full 5 axis control for the end effector, 3-axis wouldn't let you push in the right direction.

I could certainly imagine a simpler, much more rigid system, but those robots are off the shelf. Would let them focus on the software rather than spending all their time designing a 5 axis CNC machine from scratch.

I assume its slowing down at the corners to increase accuracy, not due to computing time.

 

[Comatose]

The demo in the video as shown here, pretty useless, but in theory this process could do very severe undercuts without seams or welding, and I am sure there is value to that for someone. It'd certainly justify the 6 axis robots versus a simple gantry.

 

[Strostkovy]

I think you need full 5 axis control for the end effector, 3-axis wouldn't let you push in the right direction.

I could certainly imagine a simpler, much more rigid system, but those robots are off the shelf. Would let them focus on the software rather than spending all their time designing a 5 axis CNC machine from scratch.

I assume its slowing down at the corners to increase accuracy, not due to computing time

A 3 axis system with a ball end effector can apply force in any direction. I think if this were to become dedicated machinery a dedicated 6 axis machine would be made, but I agree this is a fine and likely cheaper way to do a proof of concept.