Walter Wohlkinger from Blue Danube Robotics talks about their Airskin, a safety sensor covering robots and machines. Safety is critical in robotics and especially so in co-robotics where people work closely with robots. Walter tells us how his product, the Airskin, offers a way to add security to systems by adding a sensor to the robot.
We get to hear how the Airskin can be used to control robots by sensing the pressure on different parts of the robot. This is interesting in a co-robot context as it allows for intuitive control of robots.
Walter also talks about how the Airskin works by detecting air pressure differences inside the pad and handles puncture detection with a micro pump. Walter then shares the evolution of trying different production methods before settling on 3D-printing.
Interview: Safer human robot collaboration with a unique safety sensor with Walter Wohlkinger
Per Sjöborg, host of the Robots In Depth podcast, interviews Walter Wohlkinger to learn more about his work on Airskin and Blue Danube. Below is a transcript of the interview.
Per Sjöborg: Welcome to the podcast version of Robots in Depth and this launch episode with Walter Wohlkinger in cooperation with Wevolver. Robots in Depth is supported by Aptomica. Visit Aptomica.com to connect. You will find all past episodes and more on RobotsInDepth.com. Welcome to this episode of Robots in Depth today I'm honored to have Walter Wohlkinger here from Blue Danube Robotics and he's going to show us the Airskin which is a very good way of making your robot safe when it cooperates with people and for that matter other delicate objects I would say rather than just people. Could you tell us how this story started with this sensor?
Walter: Thank you for having me here. First the story started as it usually starts with the university spin-off. We were working an EU project and doing mobile robots and he said okay, mobile robots technologies is okay. We founded Blue Danube Robotics and we had the Airskin on the robot because we realized in moving in homes elderly you need to have a sensor skin all around the robot. This was in 2013 and in 2014 we were at Hannover fair. There we showed our robotic arms with the Airskin. Then we realized like 90% of the request was coming from the industry so maybe we are onto something. After that conference, after that fair we changed to developing Airskin as an industrial safety sensor.
Per: Could you describe the sensor and how it works?
Walter: The sensor is called Airskin and a program so we have a thin skin around some cushioning structure and in there we measure air pressure variations. You touch the sensor, you deform it, air pressure increases inside and we measure this relative pressure difference in there. To make it a safety sensor we have a small piece of electric pump in there to supply a little bit of O pressure so if you like damaged the pad, puncture, cut it then the sensor will recognize it. Turn off. Then just glue it, plug it in again and then it works.
Per: That's of course the thing that you have to have when you're measuring this air pressure. You have to know that it's air tight. This is also quite durable I understand because it is basically just rubber.
Walter: It's just rubber and the nice thing is we make it for robots, grippers, fixtures, mobile robots and all the other machines so there are like hundreds of different machines out there. How we cope with this huge variety is that we 3D print it.
Per: The physical structure we are seeing here is just 3D printed.
Walter: The full pad is 3D printed. It’s a thin air and the structure in there and we put our sensor board in there. Then we connect these pads together and we connect it to the safety IOs of the machine so where we usually connect the safety fence or the door just connect these pads.
Per: You can get them in any size or shape or form.
Walter: Yes, the smallest one is like 10 centimeters in length and the biggest one is like 60.
Per: Is that just because customers haven't wanted any other thing or is there a limitation to the technology?
Walter: There is a limitation because we measure the air pressure inside so we need a little bit of increase of air pressure that depends on the volume. The other limitation is that we usually only cover one axis of a robot and usually robot axis is about this size. We usually put like two Airskin sensors per axis so for a six, seven axis robot.
Per: Very interesting so you intend to use it in this safety mode. Is there anything else you are also thinking that it could be suitable for?
Walter: Yes so at the beginning we didn't use it as a safety sensor. We’re using it, there was one robot manufacturer he used it as a safety sensor first and then switch the mode and then we provide the system via I2C with the pressure levels so you can push away the robot. There is a user interface. It’s getting close to you. Stop it and then you push it away because you know which pad is on which axis.
Per: Very nice and since the pad is also slightly soft it would also give you that the great, that margin to have time to stop because this sensor would trigger before you hit any hard part of the robot.
Walter: Yes so it triggers really fast like 10 milliseconds and then the rest of the of the thickness we have for cushioning, for spreading out the force because the new ISO 15066 requires this maximum of impact force. It depends on your application. You send us your CAD drawing and say oh, we want to have the robot running this fast. We calculate the shape and thickness and send the pad.
Per: As they are 3D printed you can get them quite quickly. If I have a CAD file that that that I want to cover you can get the product to me quite quickly I understand.
Walter: Yes if it's a like a standard robot, easy to do. If it's a completely new robot we need a few weeks to design the pad around and usually if you order it in a few days yes.
Per: Very cool and I also understand that one of the challenges you faced when you started to develop this was that you didn't 3D print them and that made the tooling costs very expensive.
Walter: We tried every production method out there like forming the foam and then coating it but then the customer wanted a nicer surface. We tried everything and because there are lots of robot types out. More than a hundred different types of robots and we have six to seven different parts on the robot you need the production method which can cope with this variety.
Per: Now you're 3D printing it. Can you talk a little bit about the different properties of the materials I can have the sensor made in.
Walter: There are more new materials coming up but the moment this is thermoplastic polyurethane shore 90. We can make it thin and then really soft but if we want like a harder surface on the outside which is design it that way. One or two millimeters thick and have a harder surface and this comes in any color you want. There are companies in Germany they're doing all colors on it and some additional coating like dirt doesn't stick on it and sort of nano covers.
Per: Really impressive some of those to keep the robot clean and look nice and not contaminated. That brings us to for instance like food safety and all of these things because it's also washable I understand since it's just plastic.
Walter: Yes but first we're not going into the food industry. We are very taking the easiest market first assembly where people working close to robots, small electronics assembly or even the huge robots in the car manufacturing. We’re doing some sort of quality inspection on while the robot is doing some slow movement they just stopped the robot, some inspection and let it go. They really don't have any restrictions robot or on the people.
Per: The system will also be able to tell you if a pad is punctured it can tell you which pad is punctured but not exactly where I would presume.
Walter: It's per pad. If you need a higher resolution you just take more pads. The nice thing about it is it's glued under the robot. It’s double-sided tape, glue it on to the robot and if you don't need it rip it off, clean it and put a new pad on it.
Per: Really quick so minimizing downtime. Can you tell us a bit about the price of these and how we design them into our system to make them as cheap as possible because we make them easier for you to make?
Walter: The price point for a UR10. It’s about 10 to 11 pieces. You can have a gripper on there. It’s two more. It’s about it's an to more it's about €4000 to €5000 and that's about the same you pay for a fence around the robot so that's about the price point but you don't need it for the full robot maybe you only need it for one joint or for the gripper just take one or two pads.
Per: That that's more or less and this is a standard product for the UR10. You have standard products for many other robots too but say I wanted something developed. Say I had a shape that I'd need to cover what kind of price ranges are we talking about then just average kind of guidance for the customer rather than a particular price?
Walter: Per pad is €400 and depending on the size. If the size is bigger we need more and we need some hours for a 3D drawing. If it's a new robot and we can use it for us, for other customers it's cheaper. For your special racing car or whatever you want it slightly more.
Per: Also do we have standard products? I mean cube shapes or oblong so that I could just buy and put on my robot and make kind of a mosaic solution?
Walter: Yes we have one standard mini pad like 10 x 4 x 3 centimeters and you can connect them to each other and then in case a gripper like for example a suction gripper you just put on the corners four or five of these pads and connect them. For a gripper for example, this is a half part of a shank gripper. You have them there and then depending on the gripper you put it like this way or 90 degrees.
Per: You daisy chain those so that they connect to each other.
Walter: Yes up to seven can be daisy chained and then you just connect them to safety IOS.
Per: That's very interesting. You probably had prototypes out there for quite some time now. Could you tell us about any of the applications that it's been used in to advise the customer where it might be suitable?
Walter: We have it on a big PUMA robot where they have this hot glue gun. It’s a huge robot and we only cover like the first part of the robot where they do just quality inspections. If they do it right standing next to the robot but it's moving slowly doing the gluing application so this is the perfect application because usually people think of collaborative robots as ABB YuMi like small ones and doing electronics and actually you can use huge robots.
Per: Then you have to be safe of course.
Walter: Of course and another example is the CNC machine feeding so they need bigger robots to lift, to have the gripper on the robot and to lift material into the CNC machine but while they are in there the robot can do something else. You need some safety there.
Per: Has it been used in in non-manufacturing in any way? Has it been used out in the field?
Walter: Not yet because we only shipped a few prototypes out there but more and more mobile applications are, will be done so people requests like safety bumpers for their mobile robots and they need them the bumpers to be nicely designed so they want to have this this complex shape that's actually what we can do.
Per: The advantage of this sensor is the price and I also think that the very free choice you have in the shape of the sensor.
Walter: It's just the design.
Per: Also the fact that it's very durable. If you hit it hard it's not going to be affected by that. You can use it in construction industry or anything like that. It’s going to be durable. Even if it isn't it's reasonably easy to replace it also because it isn't actually built into the robot but just added to it. Thank you very much for taking the time to do an interview and it's going to be our pleasure to follow the progress as it goes along.
Walter: Thank you for the interview. You will see us at the next industrial fairs.
Per: It's going to be more shipping products.
Walter: Shipping products and you will see the product on different robots and have videos and applications there. Even on our webpage we are like trying to fill it up with applications so you see what it's used for.
Per: Perfect, thank you very much.
Walter: Thank you.
Per: I hoped you like this episode of the podcast version of Robots in Depth. This episode is produced together with Wevolver. Wevolver is a platform and community providing engineers informative content to help them innovate. It helps engineers stay cutting edge. Aptomica is the founding sponsor for Robots in Depth. Aptomica runs anything in modular robotics. Dream, rent, build. Visit Aptomica.com to connect. I am your host Per Sjöbor. Until the next episode thank you for listening.
End of transcript
The Wevolver Robots in Depth podcast is published multiple times per week. If you, or your company is interested in supporting this podcast and reach an audience of professional engineers and engineering students, please contact us at richard [at] wevolver.com