Appearance "Industrial" v/s "Collaborative" Robots

[Ox]

These new fangled cobots I guess look sleek and clean, but they appear that they would be a nightmare to have to wrench on? There is nothing exposed. The motors, and even the end effector's I/O has got to be all tucked up in there somewhere. Seems that motors could git hot if working in a warm setting? Are we not expecting cabling issues much quicker with bending/twisting much tighter?

It appears that the biggest difference between the "robot" and the "cobot" is the effort to nix pinch points.
???







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

Think Snow Eh!
Ox

 

[sfriedberg]

Pinch points, but also some of them are deliberately made weak, so when it socks you in the head it's like a punch, not an SUV impact. The ones that are not made weak have pretty serious load/torque monitoring tied into safety circuits so they will stop if they strike something at an unanticipated point in their motion (like pinning your hand between the robot, or a part it's carrying, and a machine frame).

 

[Ox]

Gosiger had a stand-alone / portable robotic tending station in both the Hardinge and the Okuma booths. Those units had a proximity detector that would put the robot in slow, or stop modes, depending on how close you were standing. Much easier to set up, but not as rigid as a light curtain. Meaning that I wonder how many times that you might trip the prox unknowingly while just being in the area and stepping in too closely?

It looked like a decent unit that was ready for service and at a price point that was lower than some other numbers that I have heard tossed around prior for similar units. It had a yellow robot with red and black motors on it.


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

Think Snow Eh!
Ox

 

[Strostkovy]

My dad used to work at a semiconductor fab. I was a well liked child by the higher ups so my dad was allowed to take spare parts for decommissioned machinery home to me. One of the machines I got was a little cartesian wafer loading robot.
The neat thing about this robot that I have seen nowhere else was that all of the joints had flat flex cables between stainless spring sheets. It was a very reliable, very slim way of getting power and signals around a moving joint. I think I still have some of the flex signals.
Anyway, it's a way cleaner approach to machine design than a bunch of loose wires shoved in a casting, but I'm sure it's more expensive.

 

[Strostkovy]

I will add that one of the big barriers to programming complex parts is the chunky joints of the robot getting in the way, and the asymmetrical joints not mirroring for left and right variants of a part. If they want it to be as easy to use as possible, then it should be as slim on the business end as possible.

What I really want is a fully symmetrical robot with through arm wire feed, an assist feeder near the end of the arm, a gimballed torch, and continuous rotation on the torch rotate joint. But that's all an engineering challenge that many would find unnecessary.

 

[thunderskunk]

The slender outside thing sounded like it was from feedback around programing locations by man-handling it into place; softer touch points for bare hands, better control, etc. I think it's just marketing, and I don't blame them. When you bring a brochure to the boss and say I want this robot for use outside a cage, it's going to sell better if it looks like a child's toy that pets kittens rather than spawn of the terminator. At the end of the day, who really uses a cobot like a cobot? At least on this forum. I just want it to move stuff after I walk away.

Productive Robotics had a booth at IMTS; fairly inexpensive cobot with their own custom joint... wobbled though. They said "we've never replaced a joint" and I laughed a bit. We ran URs into the ground, and joint replacement was a big deal. Their newer units have improved joints based on feedback to make installation easier and life longer, which made a big impression. To say "you'll never have to replace a joint" sounds like you just haven't run your machine through its life cycle. But hey maybe it does last forever, I've never worked with one of their units.

 

[TBasque]

The difference between the cobot and a robot besides the payload and in some cases the joint design is the speed. Cobots are limited to 250mm/sec where as industrial non-collaborative robots can move faster but need some protection for people. Everyone thinks I buy a cobot and use it how I please but if your cobot is moving scalpels or sharp or pointy objects then you need to have protection for the users surrounding the cobot from injury as it will not stop before hurting someone in the case of dangerous parts. A risk assesment is key in first deciding what robot to use.

As for the joints some use better parts than others. Some of them use harmonic drives or planetary gears and you will know which ones based on price. I think despite the joint gearing most industrial robots are designed to do 24x7 cycles on an assembly line. I suspect in that same space in 3 years or less (payload dependent) you would start to see issues with a cobot which I have been told by people who have bought cobots for industrial applications.

I also think that most people percieve the programming as being easier. I would say it is different as most cobots use a tablet style pendant where as robots use a more "industrial" looking pendant but they all generally have touch screens now.

Typically cobots use the grab, move, and teach strategy as the sales pitch, but if you need to do something precise or accurate you will have to use the pendant to move in tool or joint mode for anything other than rapid or unloaded moves. This is what you saw and described at the show and is the "wow" factor when demoing robots.

To cut this short, if you are running small light weight parts and you can figure out an end of arm tooling payload (yes tooling adds into the payload) setup that with your parts loaded is under the max payload (50% or less is better for longevity) then a cobot may work for you.

You will also have to figure out how to stage the rough and finished parts to keep the robot going. If you get to that then using a palatalizing routine (usually embedded in most machines) you should be 50% there. For staging, you can probably calculate how may parts you need to keep running all night.

The robot platform is the low cost part, the staging for parts, mounting, PLC or other I/O interfacing for the machine to robot and how to get the robot access to the machine (open / close doors) is usually the challenge and can be equal to or more than the robot cost.

 

[Orange Vise]

it's going to sell better if it looks like a child's toy that pets kittens rather than spawn of the terminator. At the end of the day, who really uses a cobot like a cobot? At least on this forum. I just want it to move stuff after I walk away.

Haha I'd gladly go for the Terminator but will take what I can get.

What I like about cobots is the crash protection. They're designed for human safety, but the crash detection also applies to collisions with other objects. So if the robot misgrips a part or a pallet, and then tries to load it into a vise, fixture, or ZPS incorrectly, it'll lightly bump it rather than plow through.

Makes it much faster and easier to iterate things when you don't have to worry about damage.

 

[Ox]

I know that you can turn the speeds down on a "robot".
I would ass_u_me that you can turn down the forces as well?


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

Think Snow Eh!
Ox

 

[Orange Vise]

I know that you can turn the speeds down on a "robot".
I would ass_u_me that you can turn down the forces as well?

Probably, but not sure how effective it is. My guess is that if you tune down a "rrobot" (regular robot, my term, you heard it here first) in a pick and place application, you de-rate its payload.

Cobots are designed to "crash" safely regardless of whether it's carrying 3kg or 30kg. It can carry its max rated load while still detecting a light bump.

The downside is that cobots trip out easily with otherwise simplistic tasks, like opening machine doors, where it isn't just fighting a constant force like gravity. There are workarounds but they tend to be slow/inefficient, so in these cases we use pneumatic or servo auto doors.

So I think the best use case of a cobot, in a non-collaborative environment like machine tending, is pick and place where the task may be simple but the stakes are high.

 

[Ox]

Doo you pronounce that with rolled R's as well then?
Or is that just taking things too far?


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

Think Snow Eh!
Ox

 

[Orange Vise]

The robot platform is the low cost part, the staging for parts, mounting, PLC or other I/O interfacing for the machine to robot and how to get the robot access to the machine (open / close doors) is usually the challenge and can be equal to or more than the robot cost.

Pedestals are inexpensive and can be fitted with zero-points to move cobots between machines.

Autodoors cost a bit on some machines but are easy to DIY. Haas only charges $2K for a factory autodoor and are the cheapest.

Our I/O typically consists of a just a prox switch on the door. When the door fully opens, the robot goes in and does its thing. When it's done, it physically hits cycle start. Zero invasive integration, completely humanoid.

I DIY'd the autodoor on our Mori NLX2500SY lathe. The door unlocks but doesn't open automatically on M30. So I have the robot on a timer after it hits cycle start (anywhere from 10-35 minutes depending on the cycle time). When the timer is up, the robot hits a button on a 2-way pneumatic valve that actuates a rodless cylinder that opens the door. After it does its thing, it hits the other button on the 2-way valve which closes the door, waits a few more seconds, then hits cycle start.

 

[Orange Vise]

Doo you pronounce that with rolled R's as well then?

Always.

Or is that just taking things too far?

No such thing.

 

[Ox]

Pedestals are inexpensive and can be fitted with zero-points to move cobots between machines.

Autodoors cost a bit on some machines but are easy to DIY. Haas only charges $2K for a factory autodoor and are the cheapest.

Our I/O typically consists of a just a prox switch on the door. When the door fully opens, the robot goes in and does its thing. When it's done, it physically hits cycle start. Zero invasive integration, completely humanoid.

I DIY'd the autodoor on our Mori NLX2500SY lathe. The door unlocks but doesn't open automatically on M30. So I have the robot on a timer after it hits cycle start (anywhere from 10-35 minutes depending on the cycle time). When the timer is up, the robot hits a button on a 2-way pneumatic valve that actuates a rodless cylinder that opens the door. After it does its thing, it hits the other button on the 2-way valve which closes the door, waits a few more seconds, then hits cycle start.

So, you are not monitoring a FINISHED signal at all?

So, if the machine stopped for any reason (low lube oil, low air pressure, etc) then the bot would hit the valve, but the door may not be unlocked, and so ... your bot crashes into the door and alarms out for the night? Or is this one where you have a prox on the door?

If you are not monitoring a FINISHED signal, then I Shirley wouldn't expect that you would be lookin' at a door prox.

???


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

Think Snow Eh!
Ox

 

[Orange Vise]

So, you are not monitoring a FINISHED signal at all?

So, if the machine stopped for any reason (low lube oil, low air pressure, etc) then the bot would hit the valve, but the door may not be unlocked, and so ... your bot crashes into the door and alarms out for the night? Or is this one where you have a prox on the door?

If you are not monitoring a FINISHED signal, then I Shirley wouldn't expect that you would be lookin' at a door prox.

???

On the Mori with the DIY autodoor, I now have a reed sensor on the rodless cylinder, so if the door doesn't open, the robot doesn't proceed. But yes, in the past, the robot would either try to open a locked door (fail) or just crash into the door. No biggie.

On the machines that have the factory autodoor, the robot doesn't start until the door is fully opened.

If I didn't have the ability to shrug off "crashes", I'd be much further behind and losing a lot more sleep.

 

[escapethewrmhole]

The slender outside thing sounded like it was from feedback around programing locations by man-handling it into place; softer touch points for bare hands, better control, etc. I think it's just marketing, and I don't blame them. When you bring a brochure to the boss and say I want this robot for use outside a cage, it's going to sell better if it looks like a child's toy that pets kittens rather than spawn of the terminator. At the end of the day, who really uses a cobot like a cobot? At least on this forum. I just want it to move stuff after I walk away.

Productive Robotics had a booth at IMTS; fairly inexpensive cobot with their own custom joint... wobbled though. They said "we've never replaced a joint" and I laughed a bit. We ran URs into the ground, and joint replacement was a big deal. Their newer units have improved joints based on feedback to make installation easier and life longer, which made a big impression. To say "you'll never have to replace a joint" sounds like you just haven't run your machine through its life cycle. But hey maybe it does last forever, I've never worked with one of their units.

I have one of these hooked to my Okuma. I will be paying attention to how much life I get out of it...

 

[DanielG]

Someone on PM, I forget who, used a UR5 or 10 and skipped the auto door. The robot grabbed the handle and opened the door.

 

[Orange Vise]

Played around with the Yaskawa today. It's very stout as far as cobots go. We'll likely be picking up the 20kg or 30kg version soon.

 

[thunderskunk]

Someone on PM, I forget who, used a UR5 or 10 and skipped the auto door. The robot grabbed the handle and opened the door.

Robotiq (pronounced like boutique) makes this kit for CNC load ops. It's one of those... "I mean I guess you could do it that way, but why?" Strapping light sensors to the stacklight is odd to me. A non-technical leader seeing this at a trade show would fall in love with the simplicity. It's just... why? Using a $50k robot to do the job of a $2k actuator, or even worse push a button that really aught to be $2.00 worth of wire... that's cycle time and joint wear your sacrificing. If you don't have the technical resources to do these things, why are you running CNC machines unattended? Not saying there aren't good reasons, but I'm not sure what they are myself.

 

[escapethewrmhole]

Robotiq (pronounced like boutique) makes this kit for CNC load ops. It's one of those... "I mean I guess you could do it that way, but why?" Strapping light sensors to the stacklight is odd to me. A non-technical leader seeing this at a trade show would fall in love with the simplicity. It's just... why? Using a $50k robot to do the job of a $2k actuator, or even worse push a button that really aught to be $2.00 worth of wire... that's cycle time and joint wear your sacrificing. If you don't have the technical resources to do these things, why are you running CNC machines unattended? Not saying there aren't good reasons, but I'm not sure what they are myself.

The reasons would be cost - to get a robot request and robot answer signal from the CNC you would need a robot interface and naturally the PLC logic for it from the builder or custom written and this assumes its available i.e. your machine is new enough.

The entire product Robotiq is offering is for those who do not have the robot interface and want to get in at a lower price point. Monitoring the light tree, and solenoid actuating a cycle start button is surprisingly effective and cheap.

Even to use a GPIO box and interface the robot directly to a cycle start button isn't straight forward, depending on the machine they often use a very low voltage signal that if you make the length of the cable too long it will change the resistance enough to not function, and further some machines have a dead man switch that needs both to be pressed at the same time.

I think there's a market for their product, but it's not the "right" way to do it if you want all out reliability for lights out automation.