What's new
What's new

Ideas for End Effector for Gantry style 2 or 3 axis loader - Brainstorming stage

rollerman13

Plastic
Joined
Feb 1, 2017
Good Morning Guys,

I very much enjoy reading through this forum for all sorts of information regarding manufacturing. It is such a great resource and I want to thank everyone that contributes for their help in creating such a great resource to the manufacturing world.

I oversee the manufacturing and engineering at a small company - roughly 40 employees - and I have been recently reading up on automation in high mix, low volume type operations. Obviously, these can be tricky situations in which to leverage automation.

My background is in mechanical engineering - I have been out of school and working at this company for 6 years now so I am by no means a seasoned veteran, but wouldn't consider myself a green newbie either with respect to our niche of manufacturing. Our niche involves 95% of our work done on lathes, we do very little milling, and the milling work we do is extremely basic - think keyways on shafts, and wrench flats.

With that said, I will be spending a portion of my time in the next few months researching the viability of implementing robotics with one of our CNC lathes for production. This would be our first attempt at implementation. The parts we would make in this process are all turned from Hot Rolled Steel - mainly mild carbon. And they range from 3" diameter to 8" diameter stock size. Their lengths can range from a couple inches to let's say 40". If you're familiar with this finish of steel it is neither terribly round, straight, or finished at the nominal size (it is generally slightly larger). There is no repeatability with respect to the order in which these sizes need to be machined. We are a custom shop and today's size parts will differ from tomorrow's. I've attached a picture of our Round Stock lineup just for clarity. These are obviously cut before being loaded into the CNC Lathes.

What I'm wondering is if you guys could provide some thoughts into what means we may be able to manipulate these parts with certain types of end-effectors? What types of end-effectors might be best suited for a situation where the first part may be machined from a 3.5" diameter 14" long Rnd, while the next part may be 7.25" diameter by 25" long? Obviously the big picture of this operation has MANY moving parts and will involve a lot of work. We are absolutely willing to call in the professionals for the integration, however, I believe that a solid internal understanding of the process must be gained before moving down that path.

My initial instincts are to keep it simple with a 2 axis gantry style loading system that remains centered along the spindle axis. I haven't figured out a plan to stage the material prior to the picking up the parts and loading them. These parts can get very heavy, over 500 lbs. I intend to use the 80/20 rule on this project as frequently as possible as it may just not be feasible to implement certain heavy and long parts.

A search of the forum and online has definitely helped my to understand automation and robotics, I am still definitely very green in this subject (I wish I would've concentrated much more in this arena in school), but I have yet to find a similar setup. I am certain they are out there, but I also understand the value of keeping the best kept secrets as, well, secrets...

Any thoughts or pointers would be of great appreciation. Of course I'm willing to provide more details of our situation as needed; I just wanted to keep this intro (relatively) brief.

Regards,

Garrettround stock.jpg
 
Do you have enough volume per part to change out the end effector tooling for each run, or are you trying to implement some sort of job queue where you lay out a succession of different jobs and run them unattended?
 
Volume is generally very low. We are looking at lining up a job queue as you have mentioned with a succession of jobs. At least at first we would envision an operator still tending this "cell" but of course unattended production would be best case.

So to answer your question, I do not see changing out the end effector tooling as a great option.
 
A stack of radial grippers. Something like Schunk DRG 100. Five or so spaced every 4 inches.

Radial grippers with vee jaws can grip on a variety of diameters. You'll need to do a bit of math to calculate the gripping distance for any particular diameter.

Grabbing from the side, obviously. If you're only using one or two you'd use the ones nearest the chuck.

A long part you would close all of the grippers. A short part might only use one or two. Radial grippers open to a full 180 degrees, so the one ones you aren't using for a short part won't interfere with your ability to pick and place.

I don't think you'd be able to find a three jaw with enough travel and enough gripping force to go from the end without changing out your end effector tooling, or without being extremely huge heavy and custom.

You want to make sure the grippers you use are IP67 rated (or at the VERY least 65)

I would probably have the table for blanks be a set of vee's with slots for clearance for the grippers.
 
What kind of budget are you working with? If you are wanting to automate the process, before you dump a ton of money into an X-Y gantry, consider a used or refurbished robot. You can pick up older Fanuc S420i and S430i robots 2nd hand for pennies on a dollar, and they'll run production for many years. That being said, know what you are getting into when adding a robot as far as safety etc.

As far as a gripper - sounds like you can use dual parallel grippers from PHD or Schunk spaced out accordingly, machine some v-blocks for fingers. Radial and angular grippers would also do it.
 
Schunk has cam-lock gripper arm adapters available for select gripper models. These make changing arms very, very quick if it comes to that. We use the adapters for other quick-change automation tooling also (without the gripper).

I second Irobotix post... Seek out a full 6 axis robot. It will be cheaper than a gantry and is much more versatile. Gantrys are expensive and expensive to maintain - especially expensive to maintain vs a robot when you get to trying to handle over 500 lbs. When that gantry machine is done, the gantry is most likely scrap because it can't be easily converted to other uses. The robot can easily be converted to do other jobs or machines or what have you.

At 500 lbs, you are looking at a large robot though - in Kawasaki that would be a ZX300S.
 
If humans will still be working in the cell areas along with the robots, be extremely careful with safety considerations, including contact pads on the extremities of the robot that cause instant shutdown if actuated. And you'll also want 100% surety of gripping, even a 50 pound shaft getting loose could cause a lot of damage to personnel and machinery, let along a bigger piece.

If using any axis that are driven with ball screws, include a "power off" brake to prevent unplanned movement in the event power loss. Obviously, especially for any Earth "Z" loads. I'm not sure how prone the joints in "regular" industrial robots are to back-driving, anyone here have a thought?
 
I'm not sure how prone the joints in "regular" industrial robots are to back-driving, anyone here have a thought?

All industrial robots will have axis brakes on each axis and these brakes will be stout enough to stop the max payload at the max speed in the standard required safety distance/time. Most use cycloid gearing and have no resistance to back driving. If the brakes are released without the servo powered, the arm joint will fall. I've seen 1 brake issue in 15 years of robotics operation (hundreds of robots) and it wasn't a catastrophic failure, the arm would just drift out of position 4 or 5 degrees over 24 hours of servo-off time when loaded. This was on an 11 year old bot.
 
comatose, this sounds like a nice plan. I especially like the idea of the table with blanks and the vees. I looked up the Schunk DRG 100s and they look to be a good option. The challenge of installing the part in the chuck will definitely be one of the more difficult processes to tackle as you are correct in the long jaws travel that would be needed between a 3.5" dia part and a 8" dia part. This is something I will need to spend some time considering.

Thanks for your help.

Garrett
 
irobotix, The budget question is a great question, and one in which I must do my homework on before getting too far along on this project. We are a very profitable outfit and I have not had issues with getting funds before, but I anticipate this to be the largest project in terms of cost (besides actual machine tool purchases) that I will be requesting. Since this field is so new to me I cannot quickly ballpark an estimate that would be reasonable with any confidence. I am assuming this project to run into six figures. I would love to keep it south of that level which would make it much easier to get the money needed, but I know that integration by a professional will be costly.

What is the general benefit of a parallel gripper vs. a radial gripper (or angular gripper)?

Thanks,

Garrett
 
tony, thanks for the pointer on the quick change camlock parts. We would ideally use quick change equipment wherever feasible given our need to cover a larger range of diameters and lengths for these parts.

I will have to look into the 6 axis robots more closely that you and irobotix suggested. Admittedly, they are more intimidating at first glance, but I have read great things about their versatility - and obviously they are immensely popular. I had envisioned a simple two to three axis gantry as being a LESS expensive alternative to the 6 axis bots. I'm not sure where I got that idea...but long term (and even short term) maintenance will be a large deciding factor in terms of what routes we would consider on this project.

Garrett
 
Milland, thanks for these pointers. One of my main concerns is safety. Reading through this forum it is stressed quite a lot with respect to robotic automation and I can see why. These tips will definitely come in handy as we progress through this project.
 
Guys, thank you for all the responses thus far - they have been very helpful for someone like me with almost zero experience in robotic automation. My biggest surprise has been the consensus thus far towards the six axis bots. Would it be reasonable to expect a larger footprint in the shop using a 6 axis bot? Like tony mentioned, I would expect a 6axis that can handle a payload of +500 lbs to be quite large.

My next question/concern is regarding removing the machined part from the lathe. With the large range of finish part geometries I am wondering if this will be more difficult to accomplish with the robot than even loading of the machine. As far as where the part is put after unloading - it is not critical, but gripping the part may be a challenge. Unfortunately, again, there is no rhyme or reason to the specific geometry of "today's" parts versus "tomorow's" They will all look something like these (see attachments) but they will not be repeatable, nor predictable.

I may have been naive in my thought process, however, my initial thoughts with the gantry was that a human "tending" the machine could be responsible for unloading the machine without too much issue/danger. Am I wrong in that thinking? A human inside a cell with a large 6 axis, frankly, scares the crap out of me. I would assume an automatic unloading process would be necessary which would be better in the long run anyways.

Garrettpm_j2.jpgpm_j1.jpg
 
If it were me - I'd look at a pneumatically actuated scroll chuck that would approach the turned part from the end and clamp onto the shaft, make a small rotational move in both directions looking for current on the wrist to indicate that it had a firm grip, then send the command to the lathe to release . . . of course, the moment load rating on the wrist needs to be able to handle the largest journal that you are machining.

You will also need to put the lathe in a cage with the robot or use light curtains. And a HIRA needs to be done and appropriate safety rated system designed around it.

We have integrated machines with robots tending them and depending on what your goals might be to also be able to run low volume parts with a human - it can get . . . well . . . complicated.
 
Motion, thank you for the clever thought with the scroll chuck. I will look at this option. I'm definitely concerned about the moment, but at least the parts come out of the machine (slightly) lighter than they went in :)

We will consult professionals for the safety setup for certain but I appreciate the heads up as the more I know before contacting them the better.

Garrett
 
Guys, thank you for all the responses thus far - they have been very helpful for someone like me with almost zero experience in robotic automation. My biggest surprise has been the consensus thus far towards the six axis bots. Would it be reasonable to expect a larger footprint in the shop using a 6 axis bot? Like tony mentioned, I would expect a 6axis that can handle a payload of +500 lbs to be quite large.

My next question/concern is regarding removing the machined part from the lathe. With the large range of finish part geometries I am wondering if this will be more difficult to accomplish with the robot than even loading of the machine. As far as where the part is put after unloading - it is not critical, but gripping the part may be a challenge. Unfortunately, again, there is no rhyme or reason to the specific geometry of "today's" parts versus "tomorow's" They will all look something like these (see attachments) but they will not be repeatable, nor predictable.

I may have been naive in my thought process, however, my initial thoughts with the gantry was that a human "tending" the machine could be responsible for unloading the machine without too much issue/danger. Am I wrong in that thinking? A human inside a cell with a large 6 axis, frankly, scares the crap out of me. I would assume an automatic unloading process would be necessary which would be better in the long run anyways.

GarrettView attachment 210252View attachment 210253

If you use V-block style gripper fingers, the diameters won't matter much. The center line is the same, you just need enough gripper stroke to make up for whatever the difference in the diameters is. I'm doing light parts in the 2.75" - 4.25" range, so I'm using Schunk PSH-22/1 grippers which have a 1.02" per jaw stroke, so I don't even have to change gripper arms.

One more thing you can do if you go the two parallel gripper route is to mount the front gripper on a sliding rail. (Say a INA TKVD style wide linear rail) with clamp so that you can adjust the distance from one gripper to the other at change over as part of the set-up.

Balluff has some reed sensors that you can basically "teach" clamped / open position. This would allow you to know the gripper is clamped to the part over a fairly wide range of distances and you don't have to fool with the sensor every change over other than teaching it (integrator would put a button on the teach pendant or control box to do this with)
 
Hey Rollerman - looks like you've already received some great input from MotionGuru and Tony. Both of those guys are in the automation trenches every day, and based on what I've seen them post here and the few conversations I've had with them, are very, very good at what they do.

I'm an Automation guy as well - I head up new product development for the automation division of a large, specialized machine tool builder during normal business hours.

I second using a robot in this application, mostly because they are a lot simpler to execute for someone in your position. If you decide to build a gantry, the motions of loading/unloading the machine tool are much simpler (as you will likely only have a two axis system with a gripper, or a two axis system with a rotation axis) but when you factor in having to load the gantry on the opposite end, now you're looking at a three axis system, or more. By that point, you're knocking on the door of robot price anyway, and a robot is infinitely more flexible in function than a dedicated gantry.

Additionally, any homemade loader is going to have to have a way to control it, which will also have to be homemade. If you don't have any experience with PLCs, HMIs, and programming a machine, a robot is going to likely be a much simpler solution for you. Learning the robot software for what you want it to do isn't complicated. Learning how to program a PLC and debugging it is much more difficult, in my opinion.

A robot that can handle 500lbs sounds like a monster, but it's really a pretty standard size in the industry. The Fanuc R-2000 series of robots have been around for a long time, and you can pick them up used for way, way cheaper than you'll be able to build a gantry system. We're a certified Fanuc Integrator, so that's my default, but Tony usually recommends Kawasaki robots for machine tending, and I don't disagree with him - I just know and use Fanuc every day. One thing to be mindful of is floor prep. A lot of these size robots require (by the book anyway) 18 inches of pretty serious concrete underneath them. Then again, most of your machine tools probably do too.

Just about every robot controller I've seen has provisions for I/O and power for auxiliary devices (valve banks, etc.) so they can be used as the controller for an entire system. The specific way the robot and all of it's devices (including the machine tool) will communicate depends on the robot, devices, and machine tool themselves, but just about every configuration I've ever seen can be made to talk. Most robots are also going to have a teach pendant, which will serve as your HMI (Human-Machine Interface, i.e. controller) for the cell, and some control boxes have push-button interfaces built into them that can be used as a gross 'start - stop - reset' control.

Safety is the obvious concern with these: you need to meet safety standards, make the system not a nightmare to work on, and keep it from taking up half of your shop floor space. This is easier said than done if you need to pass a safety inspection from anyone. If not, you can get away with common sense items, just keep in mind that a 250kg robot will do pretty serious damage to a human body and never even notice. We call them 'man-eaters' for a reason, and you only need to get kissed once to learn your lesson. (OK, twice, if you're me.)

I'm not doubting your ability to pull this project off, but if you get into a situation where you'd like a second set of eyes, or you want some extra engineering horsepower, feel free to reach out to me. I still do some contract work on a part-time/as-needed basis (mostly because I like doing it, and it's a nice change of pace.) I've done some work for a couple of PM members in the past. Or if you want a quote to have someone do the whole thing for you (which will probably be mind-bendingly expensive) I can have one of the Apps guys at my day job contact you.

Your particular problem of low-volume, high-mix parts is one of the largest, most challenging problems in the industry right now, and as far as I know, there isn't an off the shelf solution for it anywhere. Machine shops just vary too much in their work and approach, so it's all custom stuff, or nothing.

In any event, good luck with your project, and keep us posted!
 
Tony and Johnny, thanks again for contributing to this thread. All of this information you and the rest of the guys have shared have helped me immensely in regards to understanding the scope of this challenge.

I appreciate not sugarcoating the complexities of a project of this magnitude. I've been involved in custom automation projects similar to this before but more geared towards internal software automation and it was more or less a failure due to (in my opinion) sugarcoating complexities by the programmers. 80K spent (not including my time which isn't free to the company either) , and not much to show for it.

We are going through some big changes with our company as I write this so I don't have any confirmation on budget for FY 2018 but I anticipate this to be a necessity for our company moving forward.

I would rather spend some money now (while we are profitable and ahead of the game) to determine it will NOT work than be left in the dust by the competition who figured out it will...

Johnny, I will keep your offer in mind as I spend more time reading through and understanding all of these posts. There will definitely be help from outside the company should we move forward with this initial stint in robotic integration.

You guys have now completely sold me on the idea of a 6axis robot over a gantry style setup. I'm looking a bit more into parallel style grippers, and I have a much better, more focused, plan in mind to tackle this project.

One more question I'd like to ask is regarding the lathe itself. Is there a particular make, style, type of lathe that in your guys' opinion would lend itself best to this type of loading/unloading? In general must the controls be of certain vintage or newer? We have a combination of older and newer lathes that may lend themselves well to this process - all but one are Fanuc. I think the black sheep is a Fagor 8040 (which I do not have any experience with).

Last note: I wonder has anyone tried to cradle from underneath a larger/heavier part with a robot while unloading from a lathe?

Garrett
 
One more question I'd like to ask is regarding the lathe itself. Is there a particular make, style, type of lathe that in your guys' opinion would lend itself best to this type of loading/unloading? In general must the controls be of certain vintage or newer? We have a combination of older and newer lathes that may lend themselves well to this process - all but one are Fanuc. I think the black sheep is a Fagor 8040 (which I do not have any experience with).

Last note: I wonder has anyone tried to cradle from underneath a larger/heavier part with a robot while unloading from a lathe?

Garrett
Obviously, the newer the control the better, but Fanuc offers robot interface options for most controls. If not, then you can get a PLC guy that knows Fanuc and has the software to program it for you. (Not sure about the latest stuff, but all the older Fanuc's are a PITA to program because you can't do online PMC editing, you have to program it, save it to a PCMCIA card, reboot the control, load the new PMC, let it boot again, etc rinse and repeat for every change you need to make.)

As for cradling, my opinion is that you want that chunk of steel securely gripped when you are moving it around. If it's securely gripped, it makes no difference what orientation it's carried at or inserted via. One lesson to take when you design your system. You want to minimize the influence the robot can have on the loading of the machine. This means you should probably consider some type of loading helper inside the machine. Ideally, you want to let go of the part with the robot before it is clamped by the machine. I've used tailstocks, pneumatic cylinders, fixed and retracting V- blocks, etc. This would be true of a gantry also.

Doing this also has the advantage of you being able to get the robot out of the machine and the door closed while the clamping sequence is taking place, significantly reducing cycle time.
 
One more question I'd like to ask is regarding the lathe itself. Is there a particular make, style, type of lathe that in your guys' opinion would lend itself best to this type of loading/unloading? In general must the controls be of certain vintage or newer? We have a combination of older and newer lathes that may lend themselves well to this process - all but one are Fanuc. I think the black sheep is a Fagor 8040 (which I do not have any experience with).

Last note: I wonder has anyone tried to cradle from underneath a larger/heavier part with a robot while unloading from a lathe?

As far as the lathes go, I'm on board with Tony - generally speaking, the newer the better. But they've been using robots to tend machines pretty much since the beginning, and it can all be figured out given some time and effort. Depending on the controller brand, it can be as simple as having an M code that talks to an external controller, and then has that external controller talk back to the controller when it's done - exactly the same as using a 4th axis positioner on a VMC. It's just that the cycle the external device is running is a lot more complicated when it's a robot.

Or you can flip the whole process over and have the robot be the 'Master' of the cell, and it just tells the lathe when to open the door, release the finished part, grip the new part, move the tailstock, etc. In my experience when there is only a robot and a machine tool, the robot is the Master. If there is a PLC in the system because of the amount of other devices (conveyors, positioners, in-process gages, etc.) that require part flow management, data tracking, etc. the PLC is always the Master.

Thinking through the exact process is important. Are you going to use a robot with a single gripper, or a dual gripper? Or more? (Which wouldn't make sense for your situation from what I can tell, but is something we do often.) What functions and sensors will you need to add to the machine tool? Likely an auto-door with switches at both ends.

Visualize, visualize, visualize. And remember that a robot can be dumber than the dumbest employee who has ever worked for you, because unless you give it the means to 'know' and tell it to 'look' it won't know that it dropped a part on the way to the lathe, or that it just crashed through the window on your machine.

Getting it wrong a bunch of times is the fastest way to learn those things though :D
 








 
Back
Top