#7 How Supply Chain Disruptions Affect Product Design

Transcript:

Adar: Our guest today is Niels Jul Jacobsen. He is the CEO of Capra Robotics. Hello, Niels

Niels Hello, Adam. Thank you for having me here.

Adar: Thank you for being here. So maybe tell us a little bit about Capra Robotics.

Niels: Yeah, Capra Robotics is a new outdoor mobile robot that is focusing on this new emerging market that could be last mile delivery, it could be cleaning in the urban area and also be logistics where you move parcels from inside factories to outside. And then we started three years ago developing this new type of robot that combined some very nice features we’ve seen from other normal traditional steering robots and also from cars and combining them into one. We actually get a vehicle that is very agile and still has a low complexity in its design and that’s, of course, is very important for us when we want to roll it out into the world. And now, after three years, we have our first robot ready for the market and are selling that now primarily in Europe and to expand that to the rest of the world.

Adar: Sounds exciting. So after three years, you’re now starting to sell and move to production.

Niels: We started production about one and a half years ago. In designing that, we got a production manager who had some experience from Denmark windmills with larger structure, but basically the same. How do you build a production line that can scale from small to big? And that means that also the production engineers, they have to be involved in the last part of the sign. Because when we started with the Capra robot we had a journey of three years and we wanted to know that the first year we wanted to have the concept correct, we wanted to have the kinematic structure of the robot. Then the next one was more on. How do we make it water tight enough? Because whenever you have an outdoor robot, of course what’s very important is that it’s actually able to survive the environment. And water and dust are big hurdles there. So, the next year, we went on some design features on how to achieve that, and in the last year was more on how do we get ready for production and how do we put safety on, because the robot has to be a safety device that basically never fails. Of course, that’s not possible. Anything outdoor would eventually break because of the environment, but when it breaks, it has to break in a safe way. That’s been our journey. So, along the way and the last part, production has to be involved to ensure that their input is also very important, especially when you assemble the robot. And it’s the journey we just started and that will prolong in the next three to five years.

Adar: Yeah. So you definitely have to take many considerations into account. Especially with the robot that should survive different circumstances and different use cases and you don’t only serve like one specific use case, but different ones, say, in the design process, how do you ensure that this robot can survive all these tough situations?

Niels: Strangely enough, the whole framework was invented by my co-founder, Mads Bendt, who has been in a wheelchair for many, many years. So he has a lot of experience in driving in electric wheelchairs there, and he came up with a novel design that would solve some of the problems he has seen in the market. Unfortunately, it was not very good for wheelchairs, but it was very, very nice for robots. So, that’s why I spoke with him three years ago and said, let’s make a robotic company. So, a lot of his experience came into this design. So how can we make it? Also, when you go outside, be robust enough for the features we already know? Then I had a background from another company I started ten years ago called Mobile Industrial Robot, but also that was for indoor logistics and then we found out that it’s the right combination of having a platform that’s able to do the basic task. It has to be able to move around and do things. And then you need to have a very nice, well defined API that makes it possible for others who came on with integrators, with new ideas to put platforms on top of the robot and combine that into a kind of new product. So this ability to have a well defined API from the start, I think is very critical, yeah. And that makes it a real robot. It makes a general purpose because you can actually put a lot of things on.

Adar: Yes. The API actually allows you to connect to different technologies, provide different capabilities, probably like the autonomous navigation and stuff like that.

Niels: Yeah, I think we have some very simple one right now. But we found there’s a lot of companies out there who actually made this very perception stack where they understand what’s going on in the environment or try to navigate around things. And we don’t have to invent all that because we can go out and buy it and think we were on Automatica in June, and then we saw at least three companies who already came with this kind of technology, kind of add on technology that you can apply, and then it’s up to how complex is the task you want to solve and based that you take the different components, put them into integrated product and you can start to sell that. And some of that part is due to this ROS that had been in the robotic market for many years, I think more than ten years ago now. We started to have a kind of standard platform on how do different devices in robots communicate together. Now we are here on what we call ROS 2, which is the more industrial version of this ROS and that means that if this perception stacks, they interface to the ROS platform, is it actually very easy to plug and play that. And I think that’s going to be interesting in the future is that you can enhance the performance of your robot by adding more perception on, some of that you can develop yourself and other ones you can actually go out and buy and based on that, the robot would be smarter.

Adar: Got it. And in terms of building the actual robots, what features are a must have, what features must be in your core product, and what features are just nice to have?

Niels: Yeah, of course. Basic robustness and stability. I think it has to be designed to be able to endure some punishment, being able to run outside. And it’s also, I think we put some effort into designing it, so if something fails, it should be easy to replace it and also we put what we call redundancy into that. That means we have a four wheel mechanism with motors on all four wheels. If one or two wheels fall out, the robot should still be able to perform its task. Perhaps, the performance will go down a bit, but it should still be able to move around, things like that. We actually manage to get a system that is almost always able to move, even though it’s going to be very restricted because we found out that anything stock out in the field is terrible. You want to get it home repaired and whatever. And having this kind of redundancy is actually pretty good. I think you call it resilient nowadays. I think it’s really over the top. How do you survive the real world environment? So that was an important step in the design of robots there.

Adar: Right. So, making the robot resilient and making it even if it has some errors and some problems, it should still be operating. So this is the important part. What do you consider as nice to have features like not critical for building the robot?

Niels: I can think that, there’s another thing. How does the robot look like? How will people around the robot understand it? How will they perceive what’s going on? So, about 15 years ago, I made some tests at universities on what size should a robot be? We started with putting some mobile robots and different boxes on top. And then we tried it out with different people. And then we found out that if the robot was more than 60 CM wide, people would feel threatened about it. So the idea was if it was 80 was like, okay, perhaps it’s too big that it becomes frightening. And I think that was really interesting. So 60 CM is above the shoulder, size smallest and 80, it’s like a bear or something like that, it becomes perhaps too frightening for us. So that one design point then, actually, don’t make it too wide, because then it becomes frightening along people. Another thing is the speed, we found out that if the robot was moving too fast, if you come down a corridor and the robot comes behind you and it’s catching up, that’s scary. So, actually the robot should move a little slower than people around it. Then you have as humans, a perception, okay, now we have a robot. It’s not dangerous because I can always run away from it. 

Adar: Psychological aspects that turns into design principles that you will have to work with because you don’t want the people to be afraid of the robot.

Niels: Yeah, and now with Capra, we also work on what we call intuitive markers. So that people will understand what’s going on. We have a front and back wheel, and when that turn, it very clearly indicates in what direction can the robot move and find out that people understand that. So, that’s a kind of indication marker that makes it clear for people now, okay. If you are approaching the robot front on, and suddenly the wheel turns to the left, okay you know it’s going in that direction. It makes much more clear for you unless you can put some light on that indicates what the intention has it seen or things like that. So, all this interaction with the robot, of course they’re nice to have, but they also needed to have because we will see the robot interact with people in the street on a daily basis, and it has to be intuitive for everyone of what’s going on.

Adar: Yeah, it’s very interesting that you have these psychological principles that define eventually like how the robot will look. And you don’t want the robot to be threatening to humans because then it will be problematic when it comes into use.

Niels: Yeah, I think it’s very important for us. I think we’ve seen some robots in the market that people get offended by or scared about and that should not be the case. Robot is a kind of advanced machine that helps us as humans and I think we should keep it that way. I think robots help us, so that’s very much in our philosophy. That’s important. But it’s also part of the sign. Make it a little round. Make it also in some way nice enough, but not too nice. I think Mads Bendt has a good expression there. That’s why the company is called Capra, which means goat. Goat is nice enough, but it’s not something you want to kick. There’s also a balance on the robot, it should not be too nice because it’s still a machine and you should still understand it as such, yeah.

Adar: Yeah, got it. You talk to me earlier about supply chain issues and ways that companies have to face them and compare also to the past. You have a lot of experience and you didn’t see these same issues. So what’s your perspective about that?

Niels: Surprisingly, all you have learned for many years you have to relearn it very fast, or else you have no chance. So, suddenly supply chain is one of the main issues if you want to scale a product. Somehow, you can foresee what will happen in the future. But what if you have to scale in effects of five or ten? How would I actually do that? So you need to be much more agile on what to do. And I think in Capra, we had to redesign the PCB boards twice in the last year in order to try to find out what tips can we actually acquire and which one do we think will be available on the market to scale. So now we are designing the PCBs on what’s available and not the other way around, what would be most optimal. So that’s one point. Another one is that you need to say, okay, how can we get things more than half a year in advance? Or something like that? How do we see the demands? And of course that makes some additional risk because you say, okay, how many batteries do I have to buy? Is it my best case or worst case? Because you’re always operating in an expanding environment so, you don’t know if you sell 100 or 200 robots that can be enough for them.

Adar: Yeah, exactly. Sp how do you anticipate this demand? How do you plan in advance when you don’t really know how many robots you’re going to sell?

Niels: I think one of them was, I think, being agile enough to change some of the parts on the robots when that’s necessary. And also with the batteries you can see what supplies are available and what deals can you make with some of them and basically buy enough and then ensure you don’t have to pay all of them upfront but get suppliers to put them in stock. I think that’s what we’re trying to do now.

Adar: Right. So like, you don’t get into liquidity problems and stuff like that?

Niels: I think we would very very fast. If you’re not very careful on what deals you can get with the supplier. And sometimes when I think we have some PLCs that we got from a vendor there and then suddenly they come back and say, but we can, cannot deliver. What you do there? And then you can pay more to get the devices and that’s more or less blackmailing. And sometimes you say, okay, you have no other chance, so we do that now. But I’m not sure that we are going to be good friends in the future based on that. So pragmatism, agility and always having a lookout for what threats will come. I think that’s the way to survive.

Adar: Interesting. And how do you, like, manage this stuff? Because you don’t really know, like, you said that you design PCBs according to what’s available. So how do you even have a poll on the market and understand what kind of components will be available to you in the near future? Like, you design the robots, you don’t change design all the time. So you have to know, like, for the near future, like what kind of components will be out there. So how do you even do the forecasting?

Niels: Some of that we just go out and buy the components. We say, okay, this component is available. And then we buy a roll, two of them, a few thousand, and then we say, okay, that will be enough. And we’ll probably use that product for the next year too, so that should be okay. That’s the kind of purchase we put some capital into that, and the other ones we see, okay, how many different suppliers are available there? And we talk to them more or less directly and say, okay, what is your forecast? What do youhave in surplus for? And then based on that, you try to figure out, okay, how big a chance of this is that this component is going to increase in price. And then you also keep a lookout all the time. What’s the price level? And some components, if they become low enough in price, you just buy them in because there’s a high chance you will use them anyway. And if you come into a situation, you buy too many, you can probably sell them also. You go on a spot market to sell.

Adar: Right. And how do you like, benchmark these prices? How do you know that this price is now really low, I should buy it? Like, how do you even have that?

Niels: Oh there’s some of my electronic people who have been in that business for many years and they know, okay, how much did it used to cost two years ago? What’s acceptable? Where in the life cycle are these PCBs, or the chips are we? So, what can we expect there?

Adar: So they’re constantly in touch with the suppliers who give you custom quotes all the time, and then they say, okay, yeah I can buy it now, because now it’s a good price, and then I can reduce my rates perhaps later.

Niels Yeah, exactly. But it also means that you have people that are using a lot of time on that. So certainly that’s a new task. And normally the people, that would be the electronic guys, because it’s very hard for the purchase department to know about these things. They can get some quotation and whatever and try to press prices, but that’s really not the situation you’re in. You cannot go out and say, okay, I wanted 10% cheaper, or buy from someone else.

Adar: Yeah, have some technical experience.

Niels: Okay, you can get it from us in three years. Okay, you’re dead.

Adar: Yeah, what I hear you mentioned is like, the electronic guys are the ones who should monitor these prices, not the purchasing guys. But then there’s like some disconnect maybe between like, the engineering people and technical people are suddenly doing supply chain tasks, which is something that probably changed a little bit since the supply chain issue started. So do you have a view on how you make sure that they’re in sync with each other?

Niels: We’re not that big of a company yet. You sit together and then you say, okay, what’s going on? And then you have meetings where we put in and how do we solve that? What’s the situation, what is the risk? What component do we have coming in, and how certain are we on that? So it’s more like we call it a “war room” where we all the time try to monitor what’s going on and make people talk together and say, okay, then yes how can we solve that? So that’s the agility there. And the pragmatic one is, when do you buy anyway?

Adar: The ability to know when do you buy? Like, how does it affect the design? How does it affect your cash flow? Like, so many things that come into play when we do this supply chain challenges that you probably didn’t have to think about so much before.

Niels: Yeah, that’s completely new, I think. I talked to some guys more than 30 years ago, and they said, we have something similar sometimes there. And we just got used to a very global world where we could buy anything and it would be cheaper and could more or less arrive in a few weeks. That’s the agility now, I think, and the companies who are going to survive here, are the ones who thrive in this very uncertain enlightenment.

Adar: Incredible. Niels Jul Jacobsen CEO of Capra Robotics. Thank you for being with us today. It was fascinating.

Niels: Welcome. Thank you for letting me share our experience.