CNC Machine Automation with Robots—Laurens Wijnschenk

LAURENS WIJNSCHENK: All right. Good afternoon, everyone. As you already know since you signed up or you just joined, I'm Laurens Wijnschenk, which you can probably hear from that I'm from the other side of the ocean. I'm actually from the Netherlands. I'm here on invite by the [? HSM ?] team. They wanted me to do some classes. And this is one of them since we've been automating our CNC machines with two robotics projects we've got. One is with universal robot, a really small one. And one is with an old [? Kugar ?] robot that we actually put a whole new drive system on to see if we can get a cheap way to automate our machines.

I thought it would be cool to actually tell everyone about it and show what we've done. So first of all, I want to talk about the goals we set for ourselves. And as you see them here, they're quite vague as they're going to be very different for anyone doing this. I mean, if you just got six spindle hours on your CNC machine today, don't expect to actually go to 24/7 production with the robot. You've got to be within certain limits that you can actually improve on. So that's why I keep them vague here. It's like higher productivity, more run hours, and more consistency. I mean, the more consistency, the less hands actually touching the product, the more consistent we can produce the same thing over and over again.

Well, higher productivity is not always just in machining something faster. In this case, someone behind the machine, he can call in sick. Most of your operators are just lazy by nature. They will just stand here, wait till the machine is done, and then like, ah, the light changed color now. So I actually start to move, whereas your robot will-- the second the machine tells it, I'm done, will start to move. So it's one of those things that you should keep in mind. It's not always-- productivity is not always on actually cutting the thing faster but also on the whole process being quicker in doing that.

This is one of my strong points on actually automating your CNC machine, is that not always cutting faster will produce more parts in this case. So I'll show it to you here. If I'm actually behind the machine, I would always tell you to go with the left version of this. The quicker you make your part in this case-- higher surface speed, higher feed rate-- will cut the part in two minutes. But the tool life will only be five parts, which is fine because the price of my insert is not that high. So it will cover it for the time I actually gain with it.

But the moment I start using a robot and I'm off to bed, while the robot does the job, I do not want to come back in 10 minutes to change the insert because I'm not there. So in that case, if I just lower the speed, it actually takes longer but I get 60 parts out of it the moment I get back in the next morning instead of having five parts. So here, we actually gain a lot by going slower.

And then we come to consistency. If you've got a robot, it will always go to the same location in your machine. I mean, repeatability kind of is their game. They do the same thing over and over again for you. Especially for a lathe setup, it's important that you push the part into the jaws with the same pressure each time. Because for the parallelity, you want the part pushed against the jaws. But if you're an operator and you push it by hand, it will always be a different force. So your part will always come out different, whereas if you do it with a robot, you'll always do exactly the same thing. So if it's wrong, it's always got to be wrong. But if it's right, it should always be right.

One thing I forgot to tell about is that it will error out in a case where, for example, your [INAUDIBLE] was kept too long, whereas your operator will just usually just take the part, put it in the machine. It fits, hit the green button. You'll have a crash or a wrong part, whereas the robot will say, I was meant to go to this location. The part was too long or too short. I didn't pick up on the part. It will error out. So you'll not get a wrong part where you would probably get it when the operator was behind the machine.

We got a couple of types of handling to actually load this machine, one we mainly use. But there is another one, is we use the product handling. You'll see these a lot with [INAUDIBLE] since the chuck actually is automatic already. So there's no need for anything else. You can just pick up your part, shove it in, tell the machine to actually close the chuck, and you're ready to go. What you're doing here is actually just replacing the operator that would normally be behind the machine. This means that the initial cost will usually be low because you just need a robot to actually do the operator's job. It does require automatic vise or chuck, which you usually have on a lathe already. But you can also make this on a milling machine, of course.

The other version we've got is this pallet handling. If you look at it this way, it will always just be the same kind of thing as a CNC machine that already has pallets, only now it's a robot that's actually doing it. One of the main benefits of this is that the accuracy is actually much better since you are always clamping it on the same location. The pallet is actually handled by the machine.

One of the setbacks is that you need some kind of way to actually move the pallet about and clamp it on your machine. Especially on five [? axis ?] machines, this has got to be quite an expensive thing to do. So usually, this is only done when you buy a new machine and a whole setup that is actually ready for this. But theoretically, you could just do afterwards.

For the product handling, we've got two options. We could go with a grid in which we place our parts so the robot can pick it up. The grid is kind of the old system here. Everyone would have a feeling on how this works and what you should do with it. It's not just the easy version. It also has some advantages that you can just fill in the size of your part. And the robot could actually calculate for you where they are located and where to grab them.

The other thing is that we've got the vision system, which actually would just have a camera and look at your part, which sounds great at first. But the main setback of this is that you cannot come between the parts with an ordinary gripper, like if the parts were touching. So where you would feel like you need strict placing on the grid version of actually placing your products, you still need to do that with the vision system. Because you just cannot have parts touching anyway.

So kind of depends on the gripper that's the size of your product if you can go with a vision system. If you do, usually people use conveyor belts bringing in new products and taking away the finished ones. So you could have a whole lot of products being done that way. So you don't need really much space but actually make a lot of products. So there are cases where it's very good. On the other hand, a lot of cases, you can just go with the grid version and be done with it.

One of the things I do need to stress is the safety. And that's one of the cases why we went with the universal robot version of it since it would actually be safe to work next year. Even in that case, I mean, it would stop the moment it hit you. This sounds great. But even that has some issues if your part has sharp edges or points. If you would walk right by and it would just stick you in the eye with a point, you would still be blind. So while the robot is safe to work next to, it's not always just safe to just say, we put the robot, pick up parts, put them in a machine. And I can just stand anywhere I want.

So even with those kind of robots, you're actually looking at using something like the things you see on the right as the floor scanners, the fences. Fences can be steel or laser. And in this case, it almost looks like the laser will cut you in half. It sounds like that. It's not the case, but you've got the fences that actually use a laser to see if you actually pass through them. And that way, they would actually stop the robot or slow it down, which is one of the things mainly people use floor scanners for, is that they have certain areas in which the robot would slow down if you get close to it. But if you get even closer, it will actually stop the robot. So just walking past it wouldn't make it go into error. But if you step too close, it will actually stop it and say, this is it-- no more movement as long as you're in that area.

I can't stress enough that if any of you will start working with robots that these are things you have to think about in advance. We did it too late, not that anyone got hurt. But if you just push it to the end, it will get harder if you didn't think about one of these things and when do we get in trouble with it.

So now, it's first video, show you how our universal robots progressed from actually getting it out of the box and making it work. It was actually my brother that just said, let's buy the robot. And he told me, you're going to make it run. And we'll see where this ends. So we got it out of the box. And we started working with it. Yeah. [INAUDIBLE]. And now it's gone.

AUDIENCE: [INAUDIBLE]

LAURENS WIJNSCHENK: Yeah. [INAUDIBLE]. [LAUGHTER] You would think so. Let me try this differently.

AUDIENCE: You're new to this computer thing, right?

LAURENS WIJNSCHENK: Yeah. Totally new at it. And now why is only on that screen? Because I just told it to duplicate my screen now. So here you see me using the robot like for the first time. We bought the robot [INAUDIBLE] with it. And actually, the trick was like, what do we make it do, all right, draw up lines and arrows and actually make it do this. Most of the videos I'll show you are on my Instagram. Because we're all instant machinists. And just like [? fit ?] girls put everything up there, your whole life is-- you can see it on Instagram. So it was cool to actually show that and have people see what we were up to. And these are just simple movements. But for [? them, ?] it was really awesome to actually have a robot that did the thing I wanted it to. So with that in mind, it was actually a great thing to actually--

AUDIENCE: How long [INAUDIBLE]?

LAURENS WIJNSCHENK: Out of the box from there was actually like, I believe, an hour. And most of that was actually due to having to put it on some kind of stand. So my brother just took his welding machine, rolled it up, welded something together. And then I started to play with it. Because I'm the kind of guy that does computers, not the kind of guy that does welding. I'm not interested. It's not my thing, will never be good at it anyway, so called my brother, make it happen. So that was, I believe, like an hour, an hour and a half.

And this was the second kind of thing. This grid was actually made for the other robot we had. But it was already in. So then we started to play with it. Like can we make it more-- you see actually the boxes from the Universal Robots still back here. So it was, I believe, the same day or the next day, you were actually doing this. You were actually using all just the programming stuff that's in the Universal Robot's controller. Later on, we thought, it works. But we can improve on that and make it easier to run differently.

But here, it's starting to look like actually something loading a machine. For a milling machine, it would actually be quite similar to this. You pick it up, place it on the same kind of way just in your machine and be done with it. Luckily, these videos are all about a minute.

This is one of the things I was trying just for the fun of it. You can actually have the robot use force to sense if something is there and then actually make something happen. So it sensed with force if the part was there, then picked it up. Not really sure why this video has just screwed up.

So when we were really planning on using this robot, it was actually going to be with a lathe. So what we kind of did here is mimic a chuck that we could actually put the part into and then move away again. You would really think that I'm new to this computer stuff.

Here, you are already see that we put in some smart stuff to actually make it work more clearly. Because here, these are the pieces that were already picked up and put back. And they're not actually in the corner like it was before. Most systems you actually buy still place the product on the same place where it picked it up. The problem is that if it's moved anywhere in your system, like in the vise or during putting it in the vise, it moved like half a millimeter and will put it straight onto your plate, which will probably air out your robot.

So what you actually do it just move it out of the actual [INAUDIBLE]. Makes you notice that those parts are done as well, because sometimes you're just re-machining parts and not really clear which ones are done or not. But you can see quite clearly that here, we started to do stuff like that already because we ran into some of the issues that you see on stuff you buy still, too.

So here, it would be the first time it's actually on our first prototype of kind o what we call a cabinet with different drawers in here so we could actually have more stuff on the same kind of floor space. This whole thing that's up here, I've actually machined on our 5-axis machine, because we put the robot in there. And it turned out that, due to the way the Universal Robot is designed with glass joints here, it will actually run into the cabinet if we didn't put it on an angle. We put it on an angle and it ran to itself. So we had to make the whole thing higher as well. We solved that in the next version by moving this whole column to the front and putting it on an angle.

But it's one of those things you run into with the first prototype. And sometimes you just got to fix it, check if the rest of the system actually does run. So here actually moves it a lot like why we will do the moment it will hit the machine. In the back, you'll see our multitasking [INAUDIBLE] machine doing nothing at that very moment because I was busy doing this.

AUDIENCE: [INAUDIBLE]

LAURENS WIJNSCHENK: Yeah. One of the tricks, if it didn't find any part there, it just skip along. Because usually, your [INAUDIBLE] would just tell you, I didn't find anything out. But when I'm in bed actually making these parts-- I mean, I guess it's making parts. But I'm not really sure. So I want to have a system that actually goes on the moment I placed it wrong or don't have any issues. This is the first time we actually put it behind the machine, put parts in, and actually tried the whole setup with the connection we had on the machine. We made a connection actually with the machine. So it knew when the doors would be open-- there will be automatic doors on and the robot will actually tell the machine when to open or close the chuck. Now it's just putting one of the doors back.

One of the fixes you usually see is that the robot will actually get the doors from where it could actually work on. And we thought through that. Why should the robot do stuff like that when it's not necessary? You just move all the drawers in front, check if they are there. The moment that's true, start the robot and just push them back the moment you're done with the drawer and start on your next one.

And then here is the video some of you might have already seen that I put on YouTube actually having the machine do a production job for the first time, actually just us leaving it there the first time we went on a coffee break. It was pretty scary even though you knew nothing could actually go wrong. I mean, the robot would stop the moment something went wrong. But the feel that it's doing stuff actually when you're not there was pretty tricky. You also see that we changed the fingers on the gripper. Because usually the way that you have just two straight fingers, it's hard to grab onto something that is round. So we changed that, made a new system for that, even with the same gripper-- just changed the fingers and made it run like that.

You can see it's the second operation of these parts. But doesn't really matter if it's the first or the second operation. We did both, actually, with the robot this time. This puts it in a chuck with a certain force. You just put it in. Then it tells the machine to close the chuck. You wait for a signal that it's actually done that. Then the robot moves out, tells the machine, I'm done. Part is in there. You can close the doors and actually start making the part.

Now I would like to show you this is our other robot. It has been tested behind the same machine in this case, actually. This is our pretty large KUKA robot. It's able to handle 150 kilograms the moment it was in its normal state. Now we made some parts to put in between that's actually longer. Because otherwise, we couldn't reach. And according to KUKA now, actually, it should be able to do 100 kilograms.

The trick is that this actual robot was once used by Volkswagen actually in the production line for their cars. And it was sold to us without any control. So the trick is that we try to put different drives on and make it actually load parts for us. And if you look at it that way, it can be really cheap. Because buying another robot would be anywhere between 30,000 and 60,000 euros. And now we've got the whole thing for about 12,000 euros. So for us, it was a good challenge and actually getting it done.

It still has some issues. Because you're not completely sure on all the exact lengths of the robot, which usually are in the control you get with the robot the moment you buy it. So for us, it's always just trying new numbers and getting it dialed in. But for now, it seems to actually run great.

One problem with it is that we didn't think about all the fencing and stuff. So we run into the idea of using the floor scanners. But one of the main issues you will see with floor scanners is that actual coolant mist coming out of the machine will trigger it. And the moment it triggers it, your whole robot will say, ah, I've seen someone walking by. I'll just stop. So that causes you a lot of trouble. The moment you think, I can just run it at night, it doesn't really. Because it stops because of your own coolant mist. It doesn't work.

So that was kind of like my story of moving to robotics with our CNC machines. I'm pretty sure you all know the advantage of it, of having more time-- more production time for your machine. So I didn't go into that, more into what you need to think about and leave really much time here for you to ask me questions. Because there probably should be a lot of questions on why we did certain things. It's hard for me to actually name all of them. But if you've got questions, I'll be happy to answer them.

AUDIENCE: [INAUDIBLE] .

LAURENS WIJNSCHENK: Yeah.

AUDIENCE: [INAUDIBLE] . Which one do you think turned out more successfully, [INAUDIBLE]?

LAURENS WIJNSCHENK: The UR version for sure. Because we're actually at version two now and actually having plans to sell the whole thing as a working set. So you can actually just buy it. And we'll-- the [INAUDIBLE] reseller in the Netherlands will actually place [? you ?] [? buy ?] your machine, make all the connections, and have it working. And that's the advantages of having at least less fences and less trouble with it moving next to people. It's very convenient and the very small size of it. So you can just actually move it around. So for us, that is the best option, like most people will actually have parts between one and five kilograms, which the UR robot can just actually do. So the big one is it looks massive. It looks fun. But for most people, it will actually be overkill to have such kind of robot.

AUDIENCE: [INAUDIBLE] .

LAURENS WIJNSCHENK: Usually it's annoying. Because the lathe we [? put it by ?] pretty long. It's got, I believe, [? 1,300 ?] millimeters between the chuck and actual [? live ?] center. And even then, it's pretty close getting it in there with the longer part. So especially when you look at the [? huge ?] chuck that's on the robot, before you actually have anything grabbed into it, it's, I believe, 300 millimeters long from the face of the robot. But the trick is we have that chuck that can actually hold onto a part of 30 kilograms and move it around with two Gs of acceleration.

And if you have the different stroke of it, it can actually hold onto 100 kilograms, not like I'm actually thinking of doing that. Because the part you need to get to 100 kilograms would be massive. But we got this robot. And it wasn't much other options. I mean, we could buy this robot or a new one for much more. Yeah?

AUDIENCE: How easy was it to interface the UR [INAUDIBLE]?

LAURENS WIJNSCHENK: The trick is that your machine would need some kind of interface for it. Some machines are shipped with it. Others are not. But actually, what we did was just have just a normal pulse of [INAUDIBLE] I'm now inside of the machine or I'm outside of the machine. Close the door. Close to chuck. So I believe we need to input like 10 inputs and outputs. And that way, we can control everything. So it's not like you need fancy ways to communicate these things. You just need an input and an output board. And you should be fine.

The UR has got 16 inputs you can actually use as a safety board and then 16 are just normal inputs and the same for the outputs. So with that, it was for UR, it was pretty easy. The other robot, of course, is more like a normal [? milling ?] machine. So you've got a whole cabinet full of these boards and drives. So for that, it's a little more tricky to actually get it working. Yeah?

AUDIENCE: [INAUDIBLE] ongoing operating cost for [? maintenance? ?]

LAURENS WIJNSCHENK: From what we've seen up until now, there shouldn't be much. I mean, these robots usually are made to work 10 years straight, 24/7, go for it.

AUDIENCE: Even the UR?

LAURENS WIJNSCHENK: Even the UR-- I believe the UR is set for 20,000 hours of straight work. And if you look at it from the way we use such a robot, it usually is just waiting for the machine to do something and then grab the part, put it back. And so it's mainly just standing there waiting. One thing you would need to do is put it on its break. The UR will do that automatically if it is not moving. But other robot systems will tend to just keep all the drives on and stay like that. And you just got to tell it, put on the brake because you're waiting.

But for us, there shouldn't be much cost relating to just running it for years. Or I mean, for a UR, it's not that very expensive. So even if you would have to replace it in like two or three years, it should theoretically still be fine. But we actually plan on using it much longer than that. Anyone here bought a robot or already has a robot to play around with? [INAUDIBLE]. All right.

AUDIENCE: [INAUDIBLE] .

LAURENS WIJNSCHENK: Yeah.

AUDIENCE: [INAUDIBLE] .

LAURENS WIJNSCHENK: If you've got a CNC machine, you [INAUDIBLE] your application. I mean, you need more than one part for sure.

AUDIENCE: [INAUDIBLE] when you decide to [INAUDIBLE]?

LAURENS WIJNSCHENK: It's not like one part. But for us, we tend to do a lot of different parts. But now it turns out that people are-- even for the prototype work we do-- they come back to us like, say, can you do at least the first production run? And so that's anywhere between five and like 200. And if it goes past 20, it will be good to just put it in the robot. Because the effort you actually need to set it up now is so, so, so damn small. I mean, you give in the actual width and length of your part and the height. And it should be good to go. So with that in mind, it should be very easy even for like 20 parts to just use the robot and go do something else yourself. Yeah?

AUDIENCE: [INAUDIBLE] How do you compensate for a part that's now gone [INAUDIBLE] something [INAUDIBLE] program it in. [INAUDIBLE].

LAURENS WIJNSCHENK: Usually, your height will be about the same. Otherwise, you got your stock way too big. So that one is easy. But your contours may be different. And then like the robotics [? ribber ?] we've got on the Universal Robot actually has a pretty big stroke. So it can move like, I believe, 80 millimeters from one side to the other. I expect my parts that at least that we move with that one to be within that size. So I should actually still be able to grab them.

If it's got really weird contours, then we might need to come up with some way of actually doing that. Might be a good example of, in some cases, you might need the pallet handling, or at least device to move around. Because you cannot actually grab on it on the second side. But for lathe work, of course, it's usually [? round ?] the first time and the second time. It could be a major difference in size. That's why you've got, on the big robot, we actually made a version that has like two grippers on it. So it picks the [? raw stock ?] with one gripper and the finished one with the other. So you can actually move the jaws around. So you can actually pick both, even if there's a major difference in size. Yeah?

AUDIENCE: Do you have any idea with UR [INAUDIBLE]?

LAURENS WIJNSCHENK: They ship it as like half-finished machinery. So it's up to you to actually put everything in place that is fully safe to be used. I know in Europe, we're working against the trick that if there's something pointy to it, it can actually still damage it, even though the robot itself shouldn't do you much harm. It would mean that if you say that any [? raw stock ?] you put in there should be belt sanded so you don't have any sharp edges anymore, you should be able to use it. But the robot itself, if you don't put anything on there, you should be able to call it safe.

AUDIENCE: [INAUDIBLE]

LAURENS WIJNSCHENK: Yeah. The robot itself-- but usually from there on, it's up to you to actually implement it in a safe way. So that's why we were actually thinking about using the floor scanner as well, so that if you get near it, it will just stop. It doesn't have to go-- it pauses. So you shouldn't really hit e-stop. Because normally if you run into a big robot and you get too close, it actually hits e-stop for you, which this doesn't. It should be able to just pause since it if it's not moving, you're walking into it yourself. It's your own fault anyway. So you can explain that. But in the end, it's still up to you to have all the rest be built onto the UR being safe.

AUDIENCE: [INAUDIBLE] ?

LAURENS WIJNSCHENK: The part size, we don't usually verify. It actually does tell you if there's anything in there or not. Most of the grippers can actually do only that. So they tell you if it's in there or not. But there are some grippers out there that actually now can also tell you if it's the correct size. But for us, it was just the height that usually causes you real trouble in your machine if the height is off. So the height, you will actually feel with the robot. Because it will move on to there and then grip it.

But if it's much too high, you will run into it. So we'll actually stop and say, I found an error. Something's wrong. And if it's actually too short, it depends on if you use a system that actually does the force touch before entering it. It could tell you it's too short. Or it will just not grab onto it. And you'll note there's an issue as well. So we're not actually checking for the size. But most cases actually are caught just by having it in there or in the robot running into the block itself.

AUDIENCE: What did you do differently before [INAUDIBLE]?

LAURENS WIJNSCHENK: Not sure if I would do much different. I mean, we changed the location of where it is in the system. But actually, the robot itself, I mean it works. You've got to think about it's only for 10 kilograms of a part. So if you put in five kilograms, you will see the robot actually bend a little. And that is the same for really strong robots as well. They do the same kind of thing. So it's always an issue, like welding robots nowadays usually are tested by lasers. So they actually know what happens in any direction there at a certain point how much they actually bend and compensate for that.

But it means that you really got to be making easy way to tell the robot how much weight it's actually grabbing onto. Because that way, it can compensate a little for you for actually hanging over. And in the beginning, I just didn't care. I would just put two kilograms in and went to work with it. And that turned out to actually miss the chuck a couple of times. And I felt like it was the robot's fault. But actually, it wasn't. I just didn't put in the correct weight that the part was. So that was one of the things that just programming-wise [INAUDIBLE] keeping track that you actually got to do that stuff.

AUDIENCE: [INAUDIBLE].

LAURENS WIJNSCHENK: Theoretically, you should be able to do it. Yeah. But it's still pretty tricky because it's a 6-axis robot. And not many of those control systems actually are capable of communicating with the probe just straight out of the box. I mean, you can make anything if you wanted to. But it's probably a lot harder than you would think up front.

AUDIENCE: [INAUDIBLE] robot that can [? tell you something there. ?] Could you also think of ways to apply that [INAUDIBLE]?

LAURENS WIJNSCHENK: I'm not actually sure if the UR has that [? ability. ?] I do not believe so. But theoretically, you should be able to do that. It would just grab onto it and then check-- hover right there until it starts to move down a little. And it knows actually by its own sensors that it's moving. So theoretically, we should be able to do that. But I don't think it's just a standard option in any control.

AUDIENCE: [INAUDIBLE] robots? Do you think it's going to catch on [INAUDIBLE]?

LAURENS WIJNSCHENK: Personally, I would expect yes. Because for yourself, for example, there is going to be parts that you need more than five of, which actually should be able to put the robot behind. I mean, you've tested the program the first-- at least the first. You put the robot to work. And you go to the next machine. So for me, it sounds like a very easy solution to a problem a lot of shops actually have. I mean, a lot of shops have older machines. But a lot of machines are just stationary, waiting for someone to put a new program. But they can't because they're finishing other parts on another machine.

So it will give you a lot more time to increase your production but also just for the sake for yourself just not putting in the parts. Because we're not good at doing the same thing 1,000 times. People just not really interested in doing stuff like that. So we will start making mistakes after a couple of parts, so whereas the robot will just do that. And we should focus on making the programs. Because even with feature recognition in [? CAD ?] software, it's still the human that actually decides if it is a good program and what we should do.

So for us, I think we should see a lot of these in machine shops real soon. A lot of people are scared for some reason of doing it. I mean, it shouldn't be the money. Because we invest in machines anywhere between $100,000 and $2 million. But we're scared of investing $30,000 into a robot. So for me, it's also the journey of doing it that made it fun. But in the end, it should actually be worth it for the company and actually for our customers in the end as well. Because it should be a lot cheaper to pay for the machine and the robot running than pay for the machine and myself behind it.

AUDIENCE: So you're saying it will help you do more better [INAUDIBLE]?

LAURENS WIJNSCHENK: Exactly that-- that's at least the idea.

AUDIENCE: Do you have a rough estimate on what [INAUDIBLE]?

LAURENS WIJNSCHENK: It saves me a lot of time personally, because I would be-- I programmed the part. And I was like, yeah, it's all done. I've done my job. It works. But now I got to put in another 50 parts. So it saves me a lot of the time that I'm putting in the parts actually. So even when we do not run it at night, you would actually see that you've got six machines and only three people, you actually get a lot more done on the same kind of day. So for us, you would see that if we're making at least 20 parts for each kind of part, you would see at least double the amount of parts come out at each day, just for the eight hour day you will work.

AUDIENCE: You may have covered this already, but how long does it take to set up the robot [INAUDIBLE]?

LAURENS WIJNSCHENK: Well, what we said is like one hour to one and a half hour. We did the first moves with it. And then the whole design of the cabinet we did, of course, took a lot more time. But actually setting it up is quite easy. Having a whole program written out and have everything go at the right time, especially with the machine, of course, takes some time and testing. So it can actually take like a day. But for us, everything was sort of within a day of actually programming with the robot and doing stuff with it.

AUDIENCE: So depending on the type of work [INAUDIBLE], I think that a lot of shops have their documentation down fairly well. [INAUDIBLE].

LAURENS WIJNSCHENK: Yeah.

AUDIENCE: [INAUDIBLE] . How do you see integrating a set of documentation? How clear would it be for someone who is completely green-- [INAUDIBLE] set up a [INAUDIBLE]. If someone has no [INAUDIBLE] robots and you want them to be able to, in addition to setting [INAUDIBLE] set up the robot?

LAURENS WIJNSCHENK: Well.

AUDIENCE: If you already have the program and all [INAUDIBLE].

LAURENS WIJNSCHENK: If everything is in there, it's got to be very easy to learn. Because the only thing we need to input is the width and the length and the height of the part. And from then on, you tell the system, is it one part or 80, for example. But that's the only boxes I actually got to pick. And in the machine, it depends on what kind of gripper you have if you can actually just tell it it's this height and compensate for it in the machine. So theoretically, you shouldn't have to do anything as an operator but put in the values that you have of your stock. And it should be good to go. So if you've got everything figured out, this should be like a very small part of setting up the whole machine.

AUDIENCE: I think about this. [? For a 3-axis ?] build [INAUDIBLE].

LAURENS WIJNSCHENK: Yeah.

AUDIENCE: I can't do this to open and close a traditional grip device. I have some pallet systems. But those require more weight than [INAUDIBLE].

LAURENS WIJNSCHENK: Yeah. You would just buy an automatic vise from-- yeah, it's just as easy as that. You just got to do that. I mean--

AUDIENCE: But it's not cheap. I mean, I have four vises. And that's going to cost--

LAURENS WIJNSCHENK: But do you need all four vises to be automatic? There's a lot of jobs you would just require one.

AUDIENCE: Perhaps just one [INAUDIBLE].

LAURENS WIJNSCHENK: So if you have the automatic pallet system, you just have one pallet with the automatic vise on there, just hook up the cables. And you should be good to go. So yes, they are expensive. But we thought of the same way, like I've got the whole table full of vises and stuff I got. And then I've got to buy it all again-- usually just one. Because it's going to run at night anyway. I mean--

AUDIENCE: [INAUDIBLE].

LAURENS WIJNSCHENK: Yeah. You wouldn't do a lot more [INAUDIBLE] than putting in four parts at the same time. But on the other hand, I'm not there. So I don't really care if it takes longer than it does when I'm behind the machine. Because it's still parts that are being made while I'm asleep. So in the end, it's still a win for me.

AUDIENCE: [INAUDIBLE].

LAURENS WIJNSCHENK: Yeah, yeah, with the Universal Robot, you just press a button on the control panel. You can move it around, which is what you would do, like at least the first time for setting up where it needs to move prior to going into your machine. Because it would just be the easiest way to just hold the button till it [INAUDIBLE]. This looks good. Move here. And then you're chuck is about here.

But that is cool to actually do the first time. You're moving with the robot and it actually does what you tell it to. But on the other hand, in the end, if I got to do that for every part I'm making, to do all the teaching stuff is going to take a lot of time, whereas if you just have a full program, they actually just have parameters you fill in and have it run. That's much easier. But the teaching is fun. And it works.

AUDIENCE: So you have [INAUDIBLE] customize the job.

LAURENS WIJNSCHENK: Yeah. Actually, you shouldn't have to customize anything for a job anymore. Because you're given the height for the part the moment it's put on the table. So actually, I know that length I need to compensate for in the chuck or on the vise as well. It works. There was just nothing on the screen.

AUDIENCE: [INAUDIBLE]

LAURENS WIJNSCHENK: So you shouldn't need to teach any of it, just the first time. Because if you set it up behind a different machine, you would actually have a different move.

AUDIENCE: [INAUDIBLE].

LAURENS WIJNSCHENK: You could essentially be doing the same thing as using a bar feeder. But a lot of the times, you buy a lathe, for example, that can actually machine parts up to 200 millimeters in diameter. But you can't use the bar feeder for that kind of stuff. So for us, this was actually easier. And usually, it's actually cheaper and better option since [? parting ?] off a part is pretty tricky if you want to do a whole lot of parts that are actually pretty big in size, so whereas you just cut them with a band saw now and you lose a pretty tricky operation in the end. Seems to be it. All right.

[APPLAUSE]