David Johan Christensen is the CEO & Co Founder at Shape Robotics, creator of the Fable robot- an educational modular robot system for teaching STEM and 21st century skills to students ranging from 3nd grade to university. Davis is a researcher and entrepreneur within the area of robotics. Based on years of research, I co-founded Shape Robotics with the vision to develop the world's best educational robotics system. His research within the area of modular and distributed robotics for playful interaction, self-reconfiguration, adaptive locomotion, education, neurorehabilitation and underwater inspection. On the practical side, he has developed control strategies for self-reconfiguration, locomotion and learning. On the theoretical side, he has explored basic issues regarding scalability and morphology design. His research interests include: modular robots, self-reconfigurable robots, self-assembly, robot locomotion, rapid robot prototyping, distributed control, self-organizing and emergent control, bio-inspired control, evolutionary algorithms, artificial neural networks, reinforcement learning, learning by demonstration, programming by building, stochastic optimization, central pattern generators, behavioral adaptation, morphological adaptation, scalability, complexity, swarm robotics, human-robot interaction, artificial intelligence and artificial life.
In this podcast, David talks about modular robotics and the Fable system.
David Johan started early in robotics, getting involved already in high school. At university he found the EU project Hydra, that introduced him to modular robotics. In the Hydra project he participated in developing, among other things, the Atron self-reconfiguring modular robotics system.
We also hear how the Fable system emerged from co-operating with Lego and how it’s used all the way from 4th grade to university level brain research.
You can find out more on www.shaperobotics.com
This podcast is part of the Wevolver network. Wevolver is a platform & community providing engineers informative content to help them innovate.
Learn more at Wevolver.com
If you are interested in sponsoring the podcast, you can contact us at richard@wevolver.com
Per Sjoborg: Welcome to the podcast version of Robots in Depth. This episodes interview is with David Johan Christensen and is produced in cooperation with Wevolver. Robots in Depth is supported by Aptomica. Visit Aptomica.com to connect. You will find all past episodes and more in RobotsInDepth.com. Welcome to this episode of Robots in Depth. Today, I’m very happy to have my friend David Johan Christensen from Shape Robotics and the Technical University of Denmark here. We are going to talk about modular robotics. You guys know that I’m crazy about modular robotics. He is actually making them but let's start at the beginning. How did you get into robotics?
David: I did my education in University of Southern Denmark. Even from high school I was extremely interested in robotics so I thought I wanted to do an education in robotics. I found the best available place which is the University of Southern Denmark for me. I started being a civil engineer in robotics. After a while I saw the opportunity to join a European project called Hydra. This was a project where I did my bachelor thesis. It was a project about making something called the Atron modules. It’s kind of a self-reconfigurable robot that can rebuild their own shape. It’s like robots that can repair themselves. Robots that can change from one morphology to another and all kinds of cool things.
Per: That was one of the first systems I found, both the Hydra project and the Atron was one of the first modular robotic systems I found and that is why I’m here today. We basically got started the same way but you from the inside and me from the outside.
David: That is true. After being a bachelor student I became a PhD student. I started continuing doing my research on developing Fable and developing the Atron modules and developing all the types of modular robotics.
Per: There was the Odin system.
David: There was Lua and Ricardo Garcia did the Odin system. I also collaborated on this. We did all the different various explorations developing modular robots for all kinds of purposes not just reconfigurable but also trying to make robots that were really easy to set up and get going with. We participated in different competitions. For example at ICRA there was this challenge which was about you are in a mars scenario, something that is broken outside on the mars base. Now you have to two hours to build a robot that can solve the problem to go fix the solar panels and put them back in place and you get points. We build robots that could participate in these competitions.
Per: Because they were flexible to set up ad hoc that you could just quickly solve a particular problem.
David: We did it both with Atron and Odin and all the systems and for robot. We tried to make all our software and our tools so that it was really quick to build a robot and also program it. We usually would be some sort of remote programming but you would need some kind of scripting language to really control.
Per: Tie them together, tie the more advanced programming together.
David: You never knew what the task was so you had to have fairly generic tools that was extremely simple to use.
Per: Which is like in the industry today. We don’t know what tomorrow is going to bring. We had to be very adaptable. What you did then for the mars scenario and I was actually honored enough to be a witness to this competition. I actually attended that ICRA. The same scenario is now application to industry very much where they have to be very agile. They have to be able to adapt. It’s not like in the old days where you made the same thing for a very long. Today it's really being quick to adapt and flexibility and exploiting every opportunity. You did that as a PhD student.
David: Then I started doing a postdoc position. I got a postdoc position doing similar things so developing software frameworks, artificial intelligence algorithm that would enable robots to learn and depending whether or not it broke down. For example walking to crawl around. You could break a leg and it would learn to move around with the broken leg that it had. Then you could sort of repair the broken leg and then it would learn to use this broken leg again. It was all without any kind of centralized brain. It is distributed controller kind of focusing on how can we build, how can we make robots learn without having a centralized brain.
Per: Very important for today's application too that we continue especially for a safety perspective that we continue to come as close to our goal with the solution that we've developed even when something breaks because we can’t build a system today that requires everything to work perfectly all the time and if it doesn’t it just explodes. There is adaption to its own degrading graceful failure is a very important thing today too of course.
David: Fall tolerance. You can handle it in different ways. You can detect the fall and try to recover your control or you can also, if you have multiple robots collaborating maybe afford to lose a robot. A robot can break and may be transported back to where it can be repaired and the rest of the robots can continue the mission.
Per: You have to accept a slightly reduced output or a slightly different output but you will still be within the acceptable so and so to speak.
David: That is the type of thinking which is behind this.
Per: Then when did the idea of Shape Robotics appear for you and why did you, how did you take the step in becoming an entrepreneur?
David: What happened was that I was at the University of Southern Denmark doing my post doc and then the opportunity came up that I could go to the Technical University of Denmark in Copenhagen a few hundred kilometers away where my colleague we're about to start a big project together with Lego doing product development and research together with the Lego Company. The same guys that are doing the minestorms that we do. I thought that was too great of an opportunity to miss out on. I jumped at the opportunity. I then spent three years doing product development with Lego with their base in Technical University of Denmark. As part of that sort of process of working with Lego and learning about how they think of how should children interact with technology and then with my background in modular robotics where we also tried to make robotics really easy for anyone then combining those two things really kind of inspired us and me to think of making new kind of modular robotic system that was really easy for everybody to program and build. That became the Fable system which we continue development for a long time. Then lately we made this company Shape Robotics which are commercializing the Fable system.
Per: Could you describe the Fable system and what you intend it to be used for?
David: Fable is a modular robotic system. It consists of modules maybe this size. It’s kind of like a toolbox of robotic modules. You have some modules which are plastic parts fairly dumb with magnetic connectors and then you have other parts which have motors. They have wireless communication and batteries inside and these kinds of things, sensors, different types of modules. The idea is that you click them together. In 30 seconds you create a robot and then you can start programming that robot depending on what kind of problem you want to solve. That is what we are doing in Shape Robotics. We are trying to build this sort of library or this toolbox of robotic modules that we can deliver and offer to schools for example.
Per: What kind of age groups are you intending this to be used for?
David: We have tested it down to what is called 2nd grade in Denmark which is around eight to nine years old and then all the way up to university level with many steps in between. You can do very basic, simple programming and games in the lower grades. You can add more and more advanced programming as you progress from smaller children to the older children. The other extreme is research projects where for example one of my colleagues is working in what is called a human brain project and she is using the Fable system as a validation tool or validation platform for testing her models of the brain that they're developing within this huge European project called the Human Brain Project. You can use it from second grade to the Brain Project.
Per: That is quite the spectrum.
David: That is a big span.
Per: This also means that by extension the kids in that second they are using the same tool. They are now on a trajectory to attack their own problem of the same magnitude as the human brain because they can learn about the system and they can build on this learning. Imagine if this partner you have in the Brain Project could have started her project when she was in second grade. These kids are now 7, 8, 9 or 10 doing, what are they doing when they're 20 because they have this amazing tool.
David: I have seen kids in the schools who are working with these types of technologies 3D printing, programming robots, all kinds of things for years. It’s impressive what they can do. It’s better than what I see entering the universities most of the time they can do a lot of things. It’s a new language they learn, programing, speaking technology, thinking about innovation, thinking about how can we solve a problem for a user. How can we make this relevant for someone else?
Per: I’m thinking you and I old geezers we have to work hard to stay ahead of those guys. We are going to be passed by 11 year olds.
David: That is the reality.
Per: Isn’t that wonderful that we can give kids this opportunity?
David: It's amazing. It’s amazing to see kids work with these technologies.
Per: That is why I’m doing what I’m doing too not only for kids. I want to put this in the hands of everyone but the kids are so, they just throw themselves into this.
David: They are very immediate and they are very open to learn new things.
Per: Yes but I have actually seen the same thing when it comes to modular robotics, when I put modular robotics I did the same thing. I gave the system I then had which was the cubelets to a group of nine year old girls. That was a high pitched moment I can tell you. They were just throwing themselves at it building one thing after another. Then I tested the same thing on Harvard professors and I got the same reaction. Robotics is different from software because robotics physically you can snap the model together. That gets you into the area you are interested. Then you can come to the programming rather than start with the programming as you had to do in software. I think robotics and modular robotics have an advantage there.
David: We work a lot with the concept of flow and getting the user into a flow experience is extremely important for us and combining that with sort of play. You could say there is a correlation between play and flow. Flow is this kind of feeling where you forget time and place and you're just fully immersed in what you're trying to do. That is something that you experience a lot, for example building with Lego or programming. Many programmers feel the same thing. Reading a good book, this kinds of things and recreating those experience in building robots, programming robots is very powerful and that is something that we are aiming for.
Per: I think most of us have been in flow sometimes and it's just amazing what you can get done when you're there. It’s also I think most of us have experienced the opposite when simply nothing would work. I also think it's important to get kids to formulate their own problems, finding a problem that hasn’t been given to them by somebody that the teacher doesn’t even know the right answer but we have to be on this journey together of discovery. I think that robotics and systems like yours would be very good for that.
David: I can tell you about an example that we have done recently. We went to a public school in Denmark where the kids were first introduced to these robots so these were like 16 year olds. They were first introduced to our robots and then we went to a nursing home for elderly people. Many of them were severely handicapped and they talked to the staff and they talked to the elderly people about all the challenges they have. The staff doesn’t have much time. They need to help a lot of people all the time. The elderly maybe sit in a wheelchair and maybe had a stroke. They are paralyzed in one side and not allowed to have an electric wheelchair because that would be dangerous for the other patients. If they want to reach the remote to change the channel they can’t do that. They need the help of staff for everything. They are very dependent on the staff.
Per: They have lost their autonomy totally.
David: They met these children or young students, young people who saw all these problems and were just thinking about how can I solve this problem. I can build something and maybe I can build it for this thing and if I do that then maybe I could help her like this because she could pick it up so she needs a robot that can pick it up. Then we went back to the school and we spent a week developing welfare technology robots with 16 year olds, 15 year olds and by the end of the week the staff and the personnel came from the nursing home or the elderly care center. They saw the robots and it was amazing what this enthusiasm that these kids have put into it. What we really want to do with Shape Robotics is to kind of try to bridge the gap between just building something that may have sort of it's a toy version of the real thing. We really want to build eventually the module that allow you to build those things. It allows you to build that feeding robot for the elderly or allows you to build that robot that can pick up things for them or transport the wheelchair around. That would be fantastic so bridging the gap between the sort of education materials and something that allows them to create something real.
Per: Can you tell me a little bit about the status of the Fable system because you say they're bridging the gap between toy and prototype and not really being finished and the actual thing that is going to be out there every day doing their thing, slogging it out day after day, week after week. I like the bridge thing there. Where are we on the bridge? Are we seeing the bridge?
David: It's a different bridge maybe. It started as recent projects. We have done many iterations on these prototypes and they're very robust now in the sense that we can actually borrow them out to schools for a week and they work. We can get them back and do a little bit of maintenance and do it again. What we are doing now is we are putting them into production. We are putting them in a subset of the overall system what we call a minimum viable product. We are putting that in production and then we are going to gradually put the rest of the thing into production. Putting into production means preparing things for injection molding. Preparing things for pick and play electronic machines, making sure that have a lot of stress test, making sure that it doesn’t break when it gets into the hands of uses and making sure that all the tools are ready to go and really reliable. We are doing that right now. We are already starting to pre sell to school these robots. We are going to deliver those robots to those schools in October. That is the status.
Per: The presell they would be like 3D printed.
David: No, the presell we don’t deliver it yet. We delivered some and have it replaced later with the real thing.
Per: For testing purposes.
David: The presell now is just to get some revenue into the company. You could say and show proof of business for us and our investors. Then when we deliver the real thing then they will get the real product.
Per: Could you describe the components that are available? What kind of sensor? What kind of actuators?
David: You could say we have a library of passive parts. That is the type of you could say Lego breaks you could say that you build with and create different types of structures. The type of thing that you can create is very rigid and strong and robust. You can drop it on the floor. Then other types of components are a connector with Lego knobs on so you can build Lego parts on top of it, feet for the robot, these kinds of things. Then the other part which are the active part which is the one that contains electronics and are more expensive those ones, the first one we are going to launch here in autumn is what you call a joint module. It was two degree of freedom, Dynamics servo motors so you have precision feedback. You can control them very precisely and smooth and so forth.
Per: I know the Dynamics servos and they are a quality product.
David: You can basically, you have this kind of module that has two connectors and can move plus minus 90 degree and two axis you could say.
Per: That gives you a lot freedom right there.
David: You can build a walking robot or you can build for example, if a small group just have one of those they can build something that throws a ball with the Lego connect arm, a bit of Lego things and a ball in there and you can try to make it throw. They can do a lot of different things and then we are gradually going to put all those modules into production. More parts of modules but also a sensor module with IMU and ultrasonic distance sensor, a wheel module so you can drive. We have a design right now but we are completely changing it so it's going to be more powerful motors than what we have now. Then with encoder feedback so that they can do precise movements. Rotational modules which will and that is not going to be ready before the summer, next year which is going to allow you build robotic arms or build fairly big vehicles. We are just trying to gradually expand the capabilities of the system. Also on the software side with programming apps but also different apps that kind of enables the kids to have socially interactive robots that they can program themselves.
Per: When it comes to the programming I guess you also have to tailor the programming to different age groups. You’re not teaching the eight year old the same language as you're teaching the university students. Can you interact with these in different language also?
David: Yes. On the lowest level we are using Google's Blockly which is kind of just visual programming language that snaps together like a puzzle, like a puzzle bricks. We have very similar interface for smaller children and a slightly more advanced one with more math and functions and variables and stuff like that for later in public school and then on high school level maybe if they have very much experience in public school they will start to switch to Python. We have a very nice progression from Blockly to Python because the output of Blockly you can transform into Python and present it to the kids so that they can see the connection between the blocks and the Python code so that they can proactively learn Python and still combine it with blocks. For further level up we are also implementing Java interface and also MATLAB interface. For many at university level that is a lot of modelling courses and stuff like this where having an interface with MATLAB would make a lot of sense.
Per: I’m thinking what you're going to see here is the eight year old when he is 11 he is already at MATLABN and building stuff. He asks some stuff that you don’t know what it's about.
David: I sincerely hope so.
Per: The only way that you can get away from him is to hide from him. That is amazing stuff. It’s going to be available towards the end of this year?
David: Yes. As I said we are presenting it to schools in Denmark now. We are going to start shipping to customers in October. We are also putting it on sales.
Per: Also more modules is going to come online.
David: That is the progression of it. We are gradually going to add more modules. Also with the feedback we get from the customers figuring out is that the right modules or should we try another one.
Per: You hear lots of stuff coming in. I want to do this but I can’t. If people want to do their own module or are desperate enough to do that how can you interface with say I do a module with a laser scanner on it. There is a kickstarter now for a reasonably priced laser scanner. Can I then interface that with my software in your system?
David: Not immediately. It wouldn’t be out of the box possible. We have been discussing that. I don’t know if that is going to happen or not but it would be nice to have a small electronics board that you could integrate into your own projects.
Per: With APIs.
David: For sure also we're having and if it's further down the line is for example you could have a Raspberry Pi module. Whatever you could put into a Raspberry Pi it would fit into our system and it would come in a module size so you can snap it.
Per: Now we are talking really seriously cool stuff. There is also one called Netduino I think that has a really powerful GPU on there. Then you could start to do vision processing and really high super computer stuff. Ten years ago that would have been a national project. Now I can have multiple of them in my own modular robot.
David: The sky is the limit with these kinds of things. There’s a lot of makers out there. Of course we want to provide some tools for also those people to build cool stuff. Maybe we are not going to do the integration with the laser scanner or with the whatever but then the makers will and then it's going to be extremely interesting.
Per: The pi module is going to be cracked to the maker scene. I also think what we're going to do especially with your nice progression from the programming perspective and also it provides because the problem with 3D printing is that 3D printing a whole robot is very hard as I’m sure you know. You’ve been doing it for a while but here they can build the structure of the robot and they can program it at the level of depending how experienced they are and then they can 3D print that little last piece. That makes it so much more achievable for so many more.
David: That is exactly what they have been doing in schools. Danish schools, they have 3D printers many of them. They have been 3D printing connectors like for example with the feeding robot they would 3D print a small spoon at the end of the connector so that they could make the feeding robot the way they wanted it.
Per: It's amazing where robots are when 15 year olds are building. This is just so cool. We have to talk about costs a bit. This goes straight to the school because these systems are quite costly I understand.
David: Yes.
Per: If I wanted to buy a kit for myself and for my kids and it wouldn’t be a very big kit but I want to play with this. What kind of cost would I have to pay for a small kit?
David: A small kit is about 3000 Danish krone.
Per: It's not totally out of the ball game. It’s a big RC car.
David: We had to make some tough decisions on the costs developing this. We were actually planning to make it cheaper. The thing is that making it cheaper also makes the quality lower like selecting not Dynamics motors but poor quality motors.
Per: It doesn’t work in the long run.
David: The feedback that we got from schools were that we should definitely not do. They would much rather pay us a slightly higher costs and don’t deal with all of the problems that they see in some of their robots.
Per: I am of the same belief that it's better to have a quality product because otherwise we might actually discourage the kids because if it doesn’t work in the end it's not fun. It has to actually work in the end. We are really enthusiastic about seeing the system and I’m sure there are lots of adults out there that they think they'll buy it for their kids but they're going to play with it just as much. You’re going to be surprised how many people actually buy large sets of these for themselves. I’m one of the guys sitting right here. If my kids either get to touch them, I don’t know. The funny thing also is that we can build this stuff together with our children. From that perspective if I buy them me and my children can build things together and that is worth nearly anything from me. We can learn together. That is amazing. Thank you very much for taking the time to do an interview. It’s going to be really nice to follow your progress. I hope to have you back on the show soon with production pieces to show.
David: I would love to do that.
Per: Perfect, thank you.
David: Thank you Per.
Per: I hope you liked 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 is how 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.