Robotics Questions for a Keg Line

We have a rotary keg line that processes about a keg every 45 seconds. Currently the loading and palletizing is a manual process that requires operators to maneuver 165lb kegs. I've been tasked with improving the process to reduce the effort and risk involved.

Most existing solutions use a robotic arm, so I'm exploring this option and have a few questions for those with experience working with them.

1. How realistic is programming a robotic arm in house? A majority of our integration is done by me, both for savings and also because I spend a lot of time fixing problems from previous programmers.

2. Are used robots rife with problems and should I buy new? As an extension of this, are the features of newer models, e.g. a newer controller, worth the higher investment?

3. At what point is a vision system required to make a robotic process robust? Most of the examples I see do not utilize one, and seem to rely on consistency in the keg pattern on infeed pallets. It's possible they do have vision systems and my untrained eye didn't spot them. I've placed a few video examples below.

4. Lastly and subjectively, what brands would be best for cutting my teeth? Are there any to avoid? When looking at videos of other keg loading solutions, it seems like the FANUC R2000iB arms are mostly used. I am leaning towards this model because of the popularity. They also offer simulation software, so I was hoping to mock up the process to demonstrate to management the merits of such a system.

Here are a few examples that seem feasible:

https://www.youtube.com/watch?v=kUxHB003DcE
https://www.youtube.com/watch?v=4VDtNEdSWWk&t=98s
Manual Loading, auto palletizing: https://www.youtube.com/watch?v=Gi93K2bbGig
https://www.youtube.com/watch?v=JjzAnDpPC1o


Thank you very much for your time

I think I would contact them and have them come in and talk about it... even if you pick their brain for ideas, they may also have used equipment that they can retro to your application easier then you starting over, if your involved in the install that should be enough for you and they may also have training.

I would contact Lambrechts, seams like they have the Belgian market locked up, maybe they have a US partner or someone they recommend, I would contact them to see https://www.lambrechts-group.net/contact-us/

I always wanted to do a pick and place trainer and we are working on one now, it will have 3 stepper motors... it still a long ways out.

We use Kuka robots for our pallet operations. Pretty easy to program once you get the hang of it. Also have an ABB parallel style, I have less experience with that one.

I would not buy used, you don't know what condition they are truly in.
also, a weight of 165 lb rounded up to 200 lb for safety is a unit that not everyone needs.

I cannot comment in regards to programming the robot, I do not know your
method of packing.
I suggest to get robot vendors in to see your application and discuss your requirements. One thing you MUST have is a barrier fence for safety. I cannot cite the rules since I know they changed since I dealt with them.
james

1. We do all the programming in house at our facility, we have programmers who understand how to set it up. You could probably do it, with the correct manuals and resources, but if you are pressed for time, hire a robot integrator.

2. I would recommend talking with robot vendors and see if they will still provide support for you if you purchase a used one. Some companies will not support your robot if you by used off the market, such as Fanuc. Otherwise, you could request the person you are buying from to send you video footage of the robot actually functioning.

3. We have over 20 robots at our facility, none using vision system. We rely on sturdiness of fixture and have big enough grippers to grab onto a part, in-case our pallets are off by a few centimeters. However, having a vision system installed on your robot will make your process more robust and precise overall; but you can definitely do without if you can guarantee your keg will be in same position every time.

4. We use only fanuc at our facility. Fanuc support is excellent, top tier. They will send you many manuals and provide over-the-phone support during regular business hours. They can ship many parts overnight to you if you need (however, fanuc is known for having expensive spare parts; the robot itself is usually average price, but changing out control boards and motors are always in the thousands).

Many people use Fanuc, so I am able to find the motor/board/part I need online from ebay / other websites at much cheaper rates.

Also, I have started sending bad units to various repair shops as well (fanuc will repair your parts also, but $$$). You will most likely never run into a situation where you cannot find the part from fanuc directly or either online somewhere.

I've only worked with fanuc, so that's my experience. I'm sure other robot companies are excellent as well, but I've never used them.

1. NOT realistic if never done before; I do personally (still) consider robot programming a 'separate' discipline within the automation world.

2. You should buy new if this is the first robot you will operate.

3. You will need robot integrated vision if the rest of the system is not capable of precise positioning (+/- 1mm) or if the intended payload has variable physical characteristics; you might be able to program around variance, however, this is always time consuming and never ending.

4. Fanuc has a pretty large installed base and good reputation within the US; ABB and Kuka are on top across the pond. It really comes down to available support and documentation; capabilities are almost identical. As for simulation software, it will always require a substantial investment and I don't recall any available 'Demo's or 'grace period' free downloads.

Like others have said there are may factors that go into the type and size of robot you need, not to mention manufacture.

I would advise to buy new as well. Fanuc wont update software sometimes if the robot is not linked to your company.

I think you can get a free 30 day version of roboguide for fanuc, but if you arent doing multiple cells a year buying it probably isn't feasible since its probably $10,000. Motosim from motoman is probably $7-8000.
You can mock it up with some apple software i think for cheap to show management, if they really want it.

Workers throwing around 165lbs kegs with no lift assist is a risk for workers, insurance ect. i think thats a pretty good start to push for investment.

If you see the fanuc R2000iB used in this application thats probably the one to start looking at first. A sales rep from any robot company would gladly come in and help you with advise.

Here is some information to get you started thinking.
165lb kegs, i normally round up for the unexpected. (Say 185lbs).
Then the end of arm tooling weight..for a keg i dont know. (Factor in clamps, jaws etc. At least 50 - 100lbs.
So thats what 285lbs with no mechanical design/130Kg.
The R2000iB i think comes in a 100 or 200Kg unit.

If you want to avoid vision have accurate incoming fixturing or stop positions, being presented to the robot the same everytime, with a product as large as a keg and consistant on all sides and top doesn't seem like it would be that bad. It becomes much harder when parts can shift or have different edges and such.

I have purchased re-conditioned Fanuc robots from a Fanuc recognized partner, RobotWorx in Marion OH with success.

75% of my projects now involve one or more robots, along with a cell PLC/HMI to coordinate. Most involve new robots but a small percentage are robots re-purposed and I strongly recommend to the customer that the robot mfg come in and perform annual maintenance and a full inspection of all joints before I visit to commission it.

As far as simulation software, I've found it to be a waste of time for most applications. Using CAD you can do a cell layout and determine if the robot reach is adequate but simulations don't catch everything and it takes longer to do the simulation than the actual teaching of points. Most of the reach issues will be around the robot dressing; the cabling and hosing to operate grippers, etc, at the end-of-arm.

As far as vision, its pretty hard to beat Fanuc's vision option but I haven't used others, only Cognex or Keyence feeding vision results to a robot via the cell PLC. Vision success often comes down to the consistency of the lighting. Many vision applications have been foiled because they were near windows and as the sun changed position, the lighting conditions changed (project works great at night). And unless a bright light was supplied that increased the signal-to-noise ratio, the station ends up inside a booth (additional cost).

Just remember, you don't want to be skipping a risk assessment; when it comes to robots the level of safety that has to be provided is elevated.

Are you looking to get rid of a guy in the process or are you just trying to make the guys life easier?

Depending on the answer, you might want to look at these guys:

https://www.dalmec.com/

We have used them in the past for winding shaft manipulators. They will work with you to get custom end-of arm tooling that will mate to your load. This is not an automatic thing. It is an operator assistant that takes the load and allows the operator to more easily move it around.

Keith

Thanks for all of the answers, I really appreciate them. I think I have a clearer path forward now.

kamenges said:

Are you looking to get rid of a guy in the process or are you just trying to make the guys life easier?

Click to expand...

The operator responsibilities amount to loading empty (sometimes partially or completely full) kegs on the infeed, then manually loading filled kegs on pallets using an air assisted keg lifter. They then spray the sankey valve with sanitizer, cap it and put on a collar (branding and batch information). If a keg is underfilled, the operator must acknowledge this before the keg is discharged. All of these processes look like low hanging fruit for automation.

Despite the use of the air assisted keg lifter, there is an inherent risk involved. Pallets are double stacked requiring kegs to be lifted to chest height. I've seen operators drop kegs in a variety of ways - release the vacuum on the lifter too early, bumped into a keg already on a pallet and knocked both to the ground, used a broken pallet, etc.

For me, this is 95% about safety and 5% we could use that operator far more effectively elsewhere in our packaging operations. Other machines require much more human intervention.

There are some mixed opinions on how much of this can or should be done in house. I intend on calling FANUC and possibly robotworx on monday, but I have a few follow up questions.

• Is it necessary and advisable to have the manufacturer perform regular maintenance? We have a skilled maintenance team that works on motion machines, but are the tools and methods for maintenance so specialized as to require outside service?
  
• In regard to the programming in house, obviously this is entirely dependent on ability. I have a background in software engineering with an emphasis on computer vision and graphics, so my hope was that experience working in coordinate systems would translate. Most of my experience in automation is self taught from on the job experience and reading thoroughly, so my expectation was that robotics programming was yet another skill to learn to be more valuable. Or would you say robot programming is more of an art whose nuance is understood only after many projects?
  
• The safe route would be to try my hand before buying it. I'm hoping FANUC does offer that 30 day trial and will inquire. Are there any other ways to better understand the skills required, or should I try to get a demo from a vendor? I'm actually surprised robotics vendors don't offer free sim software to demonstrate to potential customers how they could adapt their processes.

Fanuc offers classes for programming and for maintenance - with no one at your facility with Fanuc experience its highly recommended. Programming is straight forward, "basically" a procedural language but nothing special. Understanding user & tool frames would be speeded up if you attend a class. All of the kinematics for making linear moves with multiple axes is in the brain of the robot. If you can visualize in 3D, that's half the battle.

200lbs robot seems overkill.
XY indexing table that holds the pallet.
Then just guide it off the conveyor into the same spot every time (same spot in space, but not same spot on pallet.)

My favourite vision product at the moment is IFMs O3D Scanner.
It gives you a 3d image, and because it bounces light signals off the part it seems to hold up better to different lighting conditions.

I would invite in a robot vendor to view your process and go from there. Guessing at this stage could prove costly.

If they supply reconditioned robots then great, if they don't then they don't. Saving a few $$ now is really a moot point, it's going to sit there for the next 10 to 15 years working 24/7 with little maintenenace required.

Programming in house is all very well and good, and a great challenge for a Controls guy. Once you get the hang of it, it is really not that difficult.

But you need to hone and practise your new skill or you will forget it.

If you plan on getting 20 robots over the next 2 years then go for it. If it's a one off, then just pay a guy to do it and take you through making tweaks on your own.

There will be an off the shelf solution for palletising kegs they have used before, with tweaks to the positions, it will be up and running ready for final commissioning in a day.

No point scratching around saving pennies here and there for a project that will literally last years and years.

I would go for the latest, old hardware is probably already at end of life with little support for hardware, may require continued extensive maintenance or at the very least coming close to end of life for components or support. In this day & age it appears to be the trend to end support within a 10 year period of initial manufacture or even worse replacement parts needing software modification after just months of installation due to versions, companies who supply these systems or components are often driven to update due to 1. Legislation i.e. Conformity of materials, 2. Need to keep ahead of the competitors & 3. it makes commercial sense to encourage (or even bully) customers to regularly upgrade machinery.
I don't encourage people to hijack this thread on item 3 there is enough on this subject on this site recently regarding software upgrade contracts/ versions of software versus new equipment version numbers etc. (oops have I just hijacked this thread)

Dryhops--what type of kegline are you using?

We only use half barrels and sixtels and have a vacuum type suction arm for assistiung onto a pallet.