In this episode Audrow Nash interviews Konrad Fagertun, Chief Operating Officer of nLink in Norway.They speak about a mobile robotic platform for the construction industry. The problem that they’ve chosen to solve is drilling holes in the ceiling. For a shopping mall, 400,000 holes may have to be drilled — by hand. This is a task Fagertun says “no one likes” so nLink built a robot to automate it. Fagertun discusses their robotic platform, how they chose their application, and the future plans of nLink.
Konrad has been involved with technology startups since 2006 and holds a masters degree from NTNU School of Entrepreneurship (and Boston University), with a cybernetics technology background. As one of the founders, and Chief Operating Officer, he focuses on business development, partner relations and sales in nLink’s quest to revolutionize the construction industry.
This transcript has been edited for clarity.
Audrow: Welcome to Robots Podcast. Can you introduce yourself to our listeners and talk a bit about your company?
Konrad: My name is Konrad Fagertun, I’m the cofounder of nLink Robotics. We make the world’s first mobile drilling robot for construction. In fact, I’ve just started commercializing our first robots.
We’re a small robotics company with only four fulltime engineers and several more who work by the hour. We are developing a robotics system that helps contractors do heavy work on large construction sites. The first task we started is on concrete ceiling drilling. It’s a small niche, but it is heavy and time consuming.
Audrow: Can you describe the robot and discuss more about this issue? Why construction robotics?
Konrad: With large-scale construction, for example, shopping malls or office buildings, electricians and plumbers drill thousands of holes in the ceiling before mounting constriction equipment – like cable trace, light fittings, ventilation, and air-conditioners. You have to measure every single hole and drill, which makes it very time-consuming and creates lots of noise and dust. Nobody – trust me – nobody likes this job!
We solved this by adding an industrial robot arm to a lift, then sensors and a drill. Now the robot can measure and position itself inside the room and drill with better accuracy than manual labor. It’s much faster.
Audrow: How is the user interaction? How would someone set up the robot?
Konrad: In the current version there is a laser station which is a tripod with a laser in the corner of the room. Then using an iPad, that we will supply, we made an easy to use app. You can use it manually and enter one-on-one hole or upload data from a CAD drawing. For example, you can use x and y coordinates and so forth.
Audrow: What kind of pattern does the laser tripod display so the robot knows where to drill?
Konrad: Right now it’s a laser cross in the first position to drill, or as a reference point.
Audrow: Does it create a cross or an angle? Or does it create lines of lasers where at the intersections of lines is where the machine drills?
Konrad: Somewhat. You put one laser cross where you want to start drilling or where you want the robot to look. Afterwards, the robot displays its own laser lines in the ceiling and calculates where to drill the next holes. So there’s no additional measurements it’s just getting that first hole, first light hitting, or a reference point. There may be quite a few laser lines in the ceiling, but that’s the robot trying to figure out if the ceiling or floor is tilted. It always drills perpendicular to the ceiling.
Audrow: What do you mean tilted?
Konrad: In a parking garage the floor is tilted for water allowing different floors and areas to be at alternate heights. When you drive the robot they will, in some positions, be at an angle to the ceiling. The robot will compensate and always drill perpendicular to the ceiling. There doesn’t have to be any accuracy on how to position the robot itself.
Audrow: I see. So the user puts down the laser tripod, uses an iOS or Android app to tell the robot to go, points out the spacing between the holes … is that right?
Konrad: That’s the basic function. You can also use BIM files; 3D drawings that will help subtract the holes from the drawing.
Audrow: BIM are binary files. So something you compile or build?
Konrad: Yes, it is building information modeling. These are getting popular in the US and Europe. You can subtract whole data, for example, x, y and z data – for each hole. That’s easy for the robot to interpret and drill. We even drilled a ‘smiley face’ using the information modeling. I don’t know how useful it is but does show what it’s capable of doing!
Audrow: Is it like a snowman smiley face, like a whole series of dots or is the mouth connected?
Konrad: We made a whole series of dots … 256 actually!
Audrow: Oh wow! What kind of sensors do you use to instruct the robotic arm to go to the correct location and drill the hole?
Konrad: The system uses two different sensors. One is a laser station and can be put anywhere in the room as a reference point. The second is a stereo camera system attached to the robot. This system will always look up at the ceiling and look for the laser lines it makes, then calculate how the ceiling looks if creating a model. That way it always knows where to drill and the exact height of the ceiling. It drills, for example, 1.1 inch up if that’s what want from it.
Sometimes there are restrictions in the concrete, you can’t drill too far. It’s important to control the depth of the hole. That’s very easy combining both the laser system and the camera system.
Audrow: What sensors are you using on the robotic arm so it can orientate itself? Is it the vision system or are you’re modeling the robotic arm?
Konrad: We can control each joint of the robot and calculate where it is. It’s pretty easy and something all industrial robots can do. They can tell where each of the joints of the arm is located every second. We combine with the app so the user sees any progress, like what holes have or have not been drilled. At any point you can monitor if you’re thinking, “Okay, what progress is left on this particular job? Oh, okay, we are 70% done, only a few more holes left.”
If you have any problems with a single hole the system will mark as unsolved. A worker can come back and change or fix that problem hole. There could be something in the concrete or it could be something else. There are always some irregularities.
Audrow: On the website it states this robot is constructed using off the shelf components. Can you talk a bit about that?
Konrad: We mainly use off the shelf components. We do make key components ourselves, such as the software. The largest part is the robot arm and we use a UR10 from Universal Robots, a Danish company recently bought by a US company. We also use the later laser and standard drills – standard cameras and standard vacuum cleaner and other components. We also have a patent pending system and key components that will help the system drill with increased accuracy.
Audrow: Are you also buying the lift that ‘lifts’ the robot up to the ceiling?
Konrad: That’s right. Depending on the height, we use different type of lifts. For high ceilings we need large lifts. Sometimes we need belt-driven lifts instead of wheels when there’s a difficult surface.
Audrow: Do you connect the robot to the lift, or does the operator drive the robot where it should drill the hole?
Konrad: Right now the operator has to drive the lifts.
Audrow: What should we call this particular robot? The robotic arm that drills holes in the ceiling, does it have a name?
Konrad: Actually, so far we don’t have a name. If any listeners have a good name for us we’re happy to hear it! We just call it the “mobile drilling robot” or just Drilli.
Audrow: Nice! Tell me a bit about the business model for this robot. What is the thought process behind renting it out to companies?
Konrad: Most contractors rent equipment like lifts. People are used to renting drills, excavators, all kind of machinery. They’re used to this model and they’ve asked for it so they don’t need to invest in robots. Typically, they will rent a robot for a week or two in a building, so there’s no need to own it. It’s better for them and easier. They just pay the hiring fee and save money and time right off the get go. It’s easy for all everyone, actually.
In the beginning we’ll rent it out with an operator, but eventually we want to train operators at our customers’ request. That way they can operate the robot and just rent the robots, like you rent a car.
Audrow: How large is your company?
Konrad: It’s less than ten. We’re working with large potential distributors and wholesalers who can help with distribution and logistics because these machines have to be moved from location to location in order to be effective. We will need professional logistics partners, we’re looking at Europeans companies to help with logistics in Europe and eventually worldwide.
Audrow: Does having worldwide interest make it difficult since you’re using this renting model?
Konrad: It’s a luxury problem. Right now we don’t have the capacity to fly to Hong Kong, Australia and the US at the same time. For scaling up, this is something we have to solve. In the beginning we’ll start with the two robots we have and build success stories with private customers. Then, we’ll invite people here to train and look at demos and that will make it easier for us to expand. When we want to do a job in — let’s say Brazil — we can go there for a couple of weeks, train them, then leave the robot for them to finish. That’s how we’ll solve this in the short run.
Audrow: How long is the short run?
Konrad: Probably six months, the rest of 2015 and start of 2016. It is important our own crew is with the robot to ensure the good customer care quality and the ensuring customers have no problems. We don’t want to leave customers alone with a robot that turns out to be a problem.
Audrow: Eventually will you do something similar in distribution centers at these locations that carry your robot?
Konrad: Probably, that sounds smart!
Audrow: One of your company ambitions is to revolutionize the construction industry. Can you tell me about that and why you’ve chosen this goal?
Konrad: We believe robotics is the solution to change the construction industry. Maybe in the future, the far future, we’ll see 3D printing of houses as they become more popular and feasible. However right now in the next five to 10 years I believe digital solutions and robotics will be a major part of this significant change in the construction industry.
Every time we show someone the robot they start questioning other things it can do. “Can it drill? Can it paint? Can it lay tiles? Can it do other things?” Of course it can! The important thing is positioning and navigating. When you know how to do that you can put whatever tool you want on the tip. There are possibilities in the construction industry and that’s where our hearts are right now.
Audrow: How did you choose the construction industry?
Konrad: We wanted to do something with robots and began considering industries not familiar with robotics or automation. The construction industry is a little bit behind when it comes to productivity and innovation. It seemed like a good place to start.
Audrow: How did you determine that you’re behind in terms of productivity?
Konrad: If you take a walk at the construction site you’ll see one guy with one hammer or one man with one drill working slow, carrying things. It’s heavy work. I have a lot of respect for people working there. It’s difficult.
The construction industry has also been struggling with productivity. However other industries, like the, car industry has skyrocketed in productivity. One man can produce 10 cars because robot and computer automation. There’s significant difference.
Audrow: What other industries did you consider before settling on the construction industry?
Konrad: We did consider offshore oil and gas, it’s a large industry. We also considered the medical industry and healthcare. Essentially oil, gas, healthcare, maybe education and farming, agriculture, it’s interesting and hard manual labor. Lots of opportunities!
Audrow: Definitely. How did you come on this specific issue of drilling holes in the ceiling?
Konrad: We talked to as many people as possible in the construction industry. We thought about painting and tiling first. An electricians company spoke to us and said, “Well that’s nice, but what we really need is help with measuring and drilling holes. We do this every day and its back breaking work.” They didn’t use those exact words, but that’s how we felt they said it. The idea came from an industry need.
Audrow: So then you present your robot as a solution to this problem. What has the feedback been between industry and what you’ve done with the robot?
Konrad: The feedback is pretty good so far. Everyone from workers on the floor, to CFOs, everyone loves it. The workers are happy not to continue doing a heavy workload and the CFOs see buildings completed a week ahead of schedule.
They get documentation that can be valuable for 50 to 100 years afterwards. When you do maintenance you know exactly where every important part of the building is. The documentation part has positive effects, as health and safety issues are major benefits. Plus, workers are delighted to control a robot instead of just a drill.
Audrow: How has the robot changed since you’ve introduced it to industry?
Konrad: The main form hasn’t changed much. What has changed is the way we measure how we drill in the ceiling. It’s difficult to explain. We’ve changed how we use lasers and prisms so the operator doesn’t have to do it for every single move of the robot. You measure the start and ending points where you want to drill and basically the robot calculates everything in between. There are some major savings there. Also, we have found some new solutions that will help the maintenance and service life but that’s not leasable for the operators. That’s just things we have discovered ourselves.
We’ve also seen the need for more rugged devices that can drive over heavy and difficult construction sites, especially with belts. Robots can drive over rubble and work and position it. That’s what we will learn over the next six months. We’re making a version now that will be a lot cooler!
Audrow: People that drive over rubble? Like tank treads?
Audrow: Awesome. If the robot arm and vision system is separated from the lift, and if you’re connecting the two that means the system becomes a lot more complicated. How are you going to find a way so that the robot doesn’t bump into paint cans or something laying in the way? How are you investigating? Is this a current ambition of nLink?
Konrad: Yes, we’ll do this step by step. It won’t drive around at night by itself for now but in the near future we can do that. In the beginning we’ll still need our operator to take care of it and to look after it. We’ll use crash sensors for avoiding people and obstacles. I’m not going to say easy but there are a lot of solutions we can build on so we don’t have to start from scratch.
Audrow: So if the user is always driving avoiding collisions isn’t too much of an issue?
Konrad: Yeah, exactly. Our ambitions are fully autonomous robots but that’s not going to happen this year. But it’s coming.
Audrow: Fully autonomous as in you drive a truck to the construction site and tell the robots to go and they do it?
Konrad: Yeah. Tesla currently has semi-autonomous driving so maybe you don’t have to drive down to the construction sites either.
Audrow: That would be awesome. What are some of the ambitions of nLink moving forward? You mentioned tiling and painting, what is next?
Konrad: We’ll continue with drilling robots and upgrade it to do extra tasks.
Audrow: Extra tasks as a robotic platform? And changing the end of the robotic arm so it can do other tasks?
Konrad: Yeah, we’ll add more tools. I’m not going to go into detail but it’ll be related to measuring and drilling in ceilings, walls and floors. We’ll give it extra features that will make work quicker and easier for people. Of course we’re not going to do this all by ourselves. We’re looking for partners, distributors and logistics to help. We have a couple of very interesting names on our list. This fall and winter it’s going to be very exciting for us and I hope we’ll have good news to report.
Audrow: From all of your experience with nLinks what are major challenges you’ve encountered? Have you learned anything?
Konrad: One finding is that customers think it’s amazing and they are the most important assets. We need customers to help us develop this. Not the math and algorithms but telling us how they usually work, how the process should be, the sound of concrete. Customers have told us, “Oh that’s soft concrete, you have to push a little different or, okay, the drill bit needs to be changed.” We learned a lot that would have been impossible to do in lab.
One thing we have learned is that we have to be out there with the customers. Just helping them or getting feedback. Not getting mad when they’re frustrated. That’s the first thing. Second, we learned about vision systems. There are challenges there we have overcome. 3D vision is not easy. If you can use something that’s already developed – use it! That will save you a lot of time. That’s some of the challenges we have had.
Audrow: What advice do you have for those looking to build a company around some sort of niche industry?
Konrad: We have a defined niche, that’s really the best way to do it. There’s less competition and fewer customers to talk to and the customers you find are very important in that niche. Stay in that niche as long as possible before spreading out in all kinds of directions! Find a team that can handle the technology you’re working with and the business and customer side. If you don’t understand the industry or the business you’re working in, you’ll have a hard time with the technology part.
Audrow: Thank you for being on the podcast!