Modular, flexible robotics in machining?

[Bob W.]

I have long been thinking a cell system would be the ultimate setup for my shop and I have been planning to implement one for a good long time.  Lately however, I have been thinking more and more about robotics and it they would be a better fit for what we do.  I plan to issue a PO for a Fanuc LR Mate robot with vision later this week to get on the learning curve and see what we can do in this direction.  I have two horizontals and one vertical machine so my thought is one smaller robot as seen in the first video and one larger "flying" robot on a rail above the machines as seen in the second video.  The machines would be lined up in a row and the smaller robot would load fixtures with raw material and the larger robot would place fixtures in the machines.  I currently have Schunk zero point clamping in one of my machines and I plan to buy two four-sided tombstones and implement these clamps on two sides of each.  Loading and unloading fixtures from the machines would be completely modular (consistent interface, consistent location) and that makes the trick loading the fixtures with the smaller robot, as there is a fair amount of diversity there.

 

My question is, has anyone seen robotics implemented as a flexible manufacturing system?  I know Makino has the MMCr which is basically a robotic pallet system but there is still a human loading the pallets.  It seems that robotics has typically been limited to mass production of the same part where a cell is dedicated to a family of like parts (second video) and not small volumes of diverse parts.  Any thoughts on a robotic system that would be modular and agile that could be implemented in a job shop environment?  Half of the equation is very easy because if the fixture interface is standard the larger robot has a very simple and limited task.

 

Benefits to our shop as I see it are the ability to run a job until it is complete.  Put a pallet of material in front of the robot and it runs until it is done whether it takes a few hours or a week, with zero human intervention.  This is much like a bar feeding lathe where it just goes and goes.  With the larger robot above the machines it could be parked at the end of the travel and all three run traditionally.  I'd love to hear thoughts and comments.

 

 

 

[Bob W.]

Also, the LR Mate has a repeatability of .0008" and goes for around $25k WITH VISION!  I know integration gets spendy but these are really becoming affordable.  In the second video above picture two Makino HMCs and one Makino VMC with rotary instead of the four Robodrills.  That would chew through a few pallets of material with both first AND second ops with no intervention whatsoever.  Getting a system like that up and running would be a trick though but the rewards would be huge if successful.

[jlw™]

I've seen it implemented. I've seen robots load parts as small as a little gear that got a bolt hole pattern to as large as a 900lb engine block.

 

You will have to have a full time controls guy. The vision systems are finicky at best (in my experience, ABB and Fanuc). You will have best results the less you have to change things. Get it running and leave it the hell alone!

 

Having said that, I have seen one robot run the same job for 8 years and the only problem it had was when it got hit by lightning.

 

Maybe Tyler will pop in and put in his 2 cents. In my experience, the only robot I'd consider would be a Fanuc. Dig in and check the programming too... you won't be programming it with mastercam.

[#Rekd™]

I would talk with either Michael Starry or Tyler Robertson first Bob.

 

http://www.emastercam.com/board/user/15067-starry69%E2%84%A2/

 

http://www.emastercam.com/board/user/56760-tyler-robotson/

[Rstewart]

Also, there's a robot / automation guru on practicalmachinist by the name of Tonytn36.  Just another brain to pick if interested.

[cincy k]

What's wrong with a human loading a number of pallets? Seems for the job shop environment the mix of parts is too high to justify all the end of arm tooling to the various different types of blanks.

[John@WMM]

Our shop has been an integrator of FANUC robots for over 15 years.  http://www.wauseonmachine.com/automation/automation.html 

[Bob W.]

What's wrong with a human loading a number of pallets? Seems for the job shop environment the mix of parts is too high to justify all the end of arm tooling to the various different types of blanks.

For us to get 24/7 performance out of a pallet system we would need to dedicate several pallets to a project and make several fixtures as well.  This doesn't lend itself to leaving jobs staged and after all is said and done there is a huge amount of setup and tear down which kind of goes against why I was interested in a pallet system in the first place.With just dedicating 1-2 pallets to a job the machine will only run for an hour or two after closing time at best, given the mix of parts we do.  Then there is the issue of loading the wrong stock, loading parts backwards, etc...  In my experience if it can be screwed up it WILL get screwed up sooner or later, just a matter of time.  I like the thought of a robot sitting there loading stock all day to .001" and never screwing up or loading it incorrectly but maybe this is just a pipe dream.  Robotics would be a long term direction we would take and build our processes around much like we did when moving to horizontal machines years ago.  Projects would be quoted and processes designed with robotics in mind and they would largely dictate how things were done and setup.  I don't have a crystal ball however so it is hard to anticipate the challenges or issues.  It definitely wouldn't happen in a day, or month though.  It would take over a year to get going on this.

 

Another issue is the cost.  I recently had a few cells quoted and they came back at between $850k and $1.4M.  One of these cells would allow integrating our rotary for 5-axis machining via and auto connect and the less expensive cell would not.  The hardware for the robotic system is less than $200k but the integration would be very challenging.  It would not only have the ability to serve the horizontals in 4 or 5-axis configuration, but also the vertical in 3 or 4-axis configuration.  The part loading work head issue is a big challenge to overcome though.

[bd41612]

For us to get 24/7 performance out of a pallet system we would need to dedicate several pallets to a project and make several fixtures as well.  This doesn't lend itself to leaving jobs staged and after all is said and done there is a huge amount of setup and tear down which kind of goes against why I was interested in a pallet system in the first place.With just dedicating 1-2 pallets to a job the machine will only run for an hour or two after closing time at best, given the mix of parts we do.  Then there is the issue of loading the wrong stock, loading parts backwards, etc...  In my experience if it can be screwed up it WILL get screwed up sooner or later, just a matter of time.  I like the thought of a robot sitting there loading stock all day to .001" and never screwing up or loading it incorrectly but maybe this is just a pipe dream.  Robotics would be a long term direction we would take and build our processes around much like we did when moving to horizontal machines years ago.  Projects would be quoted and processes designed with robotics in mind and they would largely dictate how things were done and setup.  I don't have a crystal ball however so it is hard to anticipate the challenges or issues.  It definitely wouldn't happen in a day, or month though.  It would take over a year to get going on this.

 

Another issue is the cost.  I recently had a few cells quoted and they came back at between $850k and $1.4M.  One of these cells would allow integrating our rotary for 5-axis machining via and auto connect and the less expensive cell would not.  The hardware for the robotic system is less than $200k but the integration would be very challenging.  It would not only have the ability to serve the horizontals in 4 or 5-axis configuration, but also the vertical in 3 or 4-axis configuration.  The part loading work head issue is a big challenge to overcome though.

 

Bob, you have a good point there.  I can envision parking a pallet of raw material next to a WSS on a Makino MMC2 and using the robot.  Remove the door on the WSS, tap into the optical pallet present recognition and use the robot to hit the pallet ready button.  I think it would be relatively easy to do for one side of a tombstone.  Multiple sides, however, could be a little tricky, you would have to mount an indexer of some kind on the WSS.  I'm sure the applications guys at Makino could tap into the MAS-A5 to detect what the robot needs to load, etc.  Just the savings of tooling up one or two pallets (instead of 5 or 6 or more) on only one job could pay for that robot and leave you with several other pallets to dedicate in the same fashion.  You could take a vacation and come back and have your mills still running.

[Sticky]

Also, the LR Mate has a repeatability of .0008" and goes for around $25k WITH VISION!  I know integration gets spendy but these are really becoming affordable.  In the second video above picture two Makino HMCs and one Makino VMC with rotary instead of the four Robodrills.  That would chew through a few pallets of material with both first AND second ops with no intervention whatsoever.  Getting a system like that up and running would be a trick though but the rewards would be huge if successful.

 

 

Where are you finding an LRmate with vision for 25K?!

 

An LRmate is a bit small for your application, I'd be looking at something more like a M20 or M710. I'm looking at both of these right now for a mill turn application, without a vision system (not needed for a part loading application imo) the M20 is around 65k usd, and the M710 around 78k usd.

 

For the type of work you are talking about it sounds like an fms is a better option. Robots work really well for single piece work flow, which if that is all you are planning on doing on the mills it can work, but it's most effective in higher volumes. For lower volume high mix work an fms will beat out single piece work flow by reduced setup times and efficiency gains from high density fixturing.

[Bob W.]

Where are you finding an LRmate with vision for 25K?!

 

An LRmate is a bit small for your application, I'd be looking at something more like a M20 or M710. I'm looking at both of these right now for a mill turn application, without a vision system (not needed for a part loading application imo) the M20 is around 65k usd, and the M710 around 78k usd.

 

For the type of work you are talking about it sounds like an fms is a better option. Robots work really well for single piece work flow, which if that is all you are planning on doing on the mills it can work, but it's most effective in higher volumes. For lower volume high mix work an fms will beat out single piece work flow by reduced setup times and efficiency gains from high density fixturing.

That was quoted new, direct from Fanuc.  My thoughts with the LR mate is it would only be loading fixtures, not the machine.  The larger robot would be shuttling fixtures to the machine and loading them.  I like the LR Mate because its speed and accuracy.  For the quoted price of the LR Mate i plan to buy one and see if we can get it to load fixtures reliably.  I'd just park it in the corner out of the way and play around with it to see what we could do.  If nothing else we could get it to deburr parts or something...

[Bob W.]

 

the M20 is around 65k usd, and the M710 around 78k usd.

 

We also got a quote for the M710 and it was in line with this.

[Bob W.]

Bob, you have a good point there.  I can envision parking a pallet of raw material next to a WSS on a Makino MMC2 and using the robot.  Remove the door on the WSS, tap into the optical pallet present recognition and use the robot to hit the pallet ready button.  I think it would be relatively easy to do for one side of a tombstone.  Multiple sides, however, could be a little tricky, you would have to mount an indexer of some kind on the WSS.  I'm sure the applications guys at Makino could tap into the MAS-A5 to detect what the robot needs to load, etc.  Just the savings of tooling up one or two pallets (instead of 5 or 6 or more) on only one job could pay for that robot and leave you with several other pallets to dedicate in the same fashion.  You could take a vacation and come back and have your mills still running.

With the overhead robot I would run the tombstones at B90 and B-90 only.  It could easily load those two sides with no indexing.

[Sticky]

I'm still shocked to here that you can get a LR200 with vision for 25k, something sounds out of whack there. Crappy vision systems usually run 5-10k all by themselves.

 

As for rotating the tombstone, get the robot to do it. It cal also open doors, press buttons and all sorts of other stuff

[bd41612]

I'm still shocked to here that you can get a LR200 with vision for 25k, something sounds out of whack there. Crappy vision systems usually run 5-10k all by themselves.

 

As for rotating the tombstone, get the robot to do it. It cal also open doors, press buttons and all sorts of other stuff

 

Unfortunately, there is a foot pedal that needs to be pressed in order for the robot to rotate the pallet in a Makino WSS.  But that could be run by a solenoid...

[bd41612]

With the overhead robot I would run the tombstones at B90 and B-90 only.  It could easily load those two sides with no indexing.

 

Sorry, I can't see the videos.  IT dept has it all locked down.

[Sticky]

Unfortunately, there is a foot pedal that needs to be pressed in order for the robot to rotate the pallet in a Makino WSS.  But that could be run by a solenoid...

 

This would be one the easiest parts of integration

 

My fms has pneumatic cylinder locks on both load stations, the foot pedal is merely a 5v i/o circuit.

[mkd]

For us to get 24/7 performance out of a pallet system we would need to dedicate several pallets to a project and make several fixtures as well.  This doesn't lend itself to leaving jobs staged and after all is said and done there is a huge amount of setup and tear down which kind of goes against why I was interested in a pallet system in the first place.With just dedicating 1-2 pallets to a job the machine will only run for an hour or two after closing time at best, given the mix of parts we do. 

possibly some pallet pool people will protest this part of your post. :-)

 

Having oodles of pallets with setups waiting in the wings is kind of the point of buying a pallet pool, so why you are concerned about tear down, I'm a little confused.

having a pool controller built by the MTB with perfect integration would be hard to beat compared to mixing robots with CNCs, IMO.

 

i'm guessing there is a REALLY good reason robotics cells are the exception, not the norm. Probably many.

 

 

 say you have 10 different part numbers. Switching between them with a pallet controller is easy as pie; you just drag and drop them into Que. Now imaging the same change over on a robotic system.... seem like it would take an act of congress to get one part number changed over, let alone 10.

 

if you hire a guy that can program and run a Hori, he can pretty much handle a pallet pool. Not so for a robot.

 

25k is not a huge investment and if it doesn't work you can always sell it.

I would love to hear a success story here, as i think it is an interesting "why not?". I hate to be a naysayer but i just keep thinking of a couple job shops with second spindle lathes. they either never used it or rarly used it holding a live center. LOL..true story.

hope the best for the project.

[mkd]

.

[mkd]

.  Then there is the issue of loading the wrong stock, loading parts backwards, etc...  In my experience if it can be screwed up it WILL get screwed up sooner or later, just a matter of time.  I like the thought of a robot sitting there loading stock all day to .001" and never screwing up or loading it incorrectly but maybe this is just a pipe dream.

good fixture design will eliminate 80%

probe cycle for the other 20%

IMHO

[Bob W.]

possibly some pallet pool people will protest this part of your post. :-)

 

Having oodles of pallets with setups waiting in the wings is kind of the point of buying a pallet pool, so why you are concerned about tear down, I'm a little confused.

 

 

The programs I run typically cycle a pallet in 1H 45M on average, then it gets kicked out and the next pallet shuttled in.  That means after I load the last piece of stock the machine would run for 3.5 hours or so unless I dedicate several pallets and fixtures to this project.  If I leave the building at 5:00 and want the machine to run until 8:00 the next morning unattended I would need to dedicate 8 pallets (and 32 fixtures) to the project.  Even if I have a few projects going there would be some duplicate setup.  I can mix in other repeat projects in with it but if I don't have a PO in hand this is a huge liability.  I wouldn't leave my machine with 8 pallets dedicated to a project once it was completed so there would be significant tear down and setup to operate this way.  With a robotic solution there would be two fixtures and a pallet of material and it would run non-stop until complete.

 

Another issue with this is I was recently informed that my existing 400mm horizontal with rotary table capability would NOT work in the cell system because they don't offer an auto connect for it which leaves me with a big dilemma.  To get into a cell I would need to buy a new machine AND the cell system, and it would be a single machine system.  This is where a robotic solution becomes attractive.  It could handle anything I put in front of it.  Makino is building this very system, the MMC-R which is a robotic cell system that will run any machine (horizontal, vertical, 3-axis, 4-axis, 5-axis) with a Pro5 control.  It doesn't have a robot loading the fixtures though.

 

It does make me a little nervous that this isn't being done but at the same time the benefits would far out weigh a traditional cell system if done well.  I guess what I am looking at is Makino's MMC-R with a robotic fixture loader on the front end. 

 

I'm thinking robotic routines filed on the server, quick change heads, grippers, etc..., and pretty standard fixture design.  Initial setup would be daunting and even if changing over took two hours it would still run until done after that.  For the same project listed above that would amount to 30+ hours of run time for 2 hours of setup and for what we do I don't think we could get better spindle utilization from a cell.

[Bob W.]

Okay, so here is a pie in the sky scenario of what could be done.  A completely automated cell where you set a pallet of material in front of the robot and finished parts come out, and not just one operation finished, COMPLETELY finished.  Here is how it would be done...

 

Tools:

Fastems CTS, a tool storage system that has a magazine capacity of thousands of tools and can serve multiple machines (I believe)

 

 

 

Fixtures:

Makino MMC-R, can load fixtures into 5-axis, 4-axis, VMC, HMC, etc...

 

 

Parts:

Robotic arm loading parts into a fixture.

 

 

All of these things have been done, and done well.  I'm just not sure if they have been done in the same cell.  This isn't really my intention but I might get there after 10 years, working on it during my spare time...  Oh well, fun to dream big.

[cincy k]

Maybe in missing something. Why are you only considering running one job at a time? You have a mmc with pallets setup for various different jobs you currently have orders for. Job A has two pallets netting 1 hour 45 minutes unattended time. Job B the same. Job C the same. Job D the same. If there aren't enough pallets in the mmc use schunk zero point tombstones in the cell. Load fixtures and go. Done. Don't need uber amounts of extra fixtures.

Makino does offer integrating 5 axis into an fms. Kme does it as well now. It CAN be done. Though I think you'd be better with a d500. I think where your current machines come up short is tool storage capacity to run multiple jobs at once.

[cincy k]

Definitely understand dreaming big. We all do it. Just seems like adding to many separate things together and asking them to work all the time is asking for constant headaches. Might take years to perfect while going broke in the mean time. Playing devils advocate.

[Bob W.]

Maybe in missing something. Why are you only considering running one job at a time? You have a mmc with pallets setup for various different jobs you currently have orders for. Job A has two pallets netting 1 hour 45 minutes unattended time. Job B the same. Job C the same. Job D the same. If there aren't enough pallets in the mmc use schunk zero point tombstones in the cell. Load fixtures and go. Done. Don't need uber amounts of extra fixtures.

Makino does offer integrating 5 axis into an fms. Kme does it as well now. It CAN be done. Though I think you'd be better with a d500. I think where your current machines come up short is tool storage capacity to run multiple jobs at once.

It seems with the workflow he typically have we will have one big project and numerous smaller projects.  A FMS would definitely make us more efficient throughout the day due to projects set up and how quickly we could change over but I think we would struggle to get more than 10-12 hours per day out of it without me standing in front of it loading parts well into the evening.  As it stands, my current A51nx with the rotary couldn't be integrated into the FMS and retain the 5-axis capability unless I was to do it manually, which wouldn't be worth it.  My goal in these posts is to get comments like this that make me sit back and think and see what might go wrong, what the challenges might be, and what I might be missing.  I am still open to anything and I have the FMS quote on my desk next to the LR Mate quote.  Decisions decisions...  Thanks for the input, I do appreciate it.