Daniel Pizzata talks about how his passion for modular robotics has driven him to start Modbot, a company building a robotics platform and creating a community. The goal of Modbot is to bring robots out of the research labs and make them accessible to everyone. Daniel started out in the defense sector in Australia using robotics to measure radio transmissions. He worked on many different projects, but felt that he wanted to work with technology that was more widely applicable in society, and follow his ambitions. Daniel also talks about how he met his co-founder Adam Ellison and how they had the idea of a platform and a community that could widen the range of people that are able to develop robotics and automation solutions. This turned out to be a life-changing journey with many intense moments and amazing experiences. Daniel then shares how living on floor 7½ in the mezzanine corridor was one of the interesting aspects of moving to San Fransisco to pursue his dream of starting Modbot.
More information on https://www.modbot.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: Welcome to the podcast version of Robots in Depth and this launch episode with Daniel Pizzata 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 Robots in Depth. Today, I’m honored to have Daniel Pizzata from Modbots here. We’re going to talk a lot about modular robotics and the great products they've got coming up. I'm so excited that we're going to see modular robotics out there in the world but we're going to start where I usually start. How did you discover the field of robotics? How come you're now building cool modular robot systems?
Daniel: My advent into robotics actually began in the defense sector in Australia. I grew up on the West Coast and ended up joining the defense sector in 2009, 2010 thereabouts, maybe earlier but I am I worked for an agency called the Defence Materiel Organisation whose primary mandate is to equip and establish all the armaments and that can include everything from aircraft carriers all the way down to M&Ms for the service sector whether you're in the maritime aerospace or land division. I actually worked throughout a variety of different areas inside of there until I ended up in the land division supporting a project team that wanted to have autonomous vehicles. It was a vehicle system that needed to probe the electromagnetic spectrum of a richly configured vehicle covered and like laden with antenna systems. The point will largely be is we would by UHF through VHF through a whole bunch of different antenna systems and they would cause cosa interference. We had to characterize the system and when you think about a military vehicle in the middle of the desert. It doesn't have a cell tower nearby. It is the cell tower. You can't be close to it when irradiating. When that energy is coming off those antenna systems it's dangerous.
You can't probe this with a human being. You need to have a vehicle that ideally isn't a communication sink and affects the EM spectrum and is also able to basically be left alone for a handful of hours to take thousands of measurements instead of getting human to do it that would cost a significant amount of money and we would end up with maybe ten measurements but the complexity of the integration of the electronics and hardware and the designs, communication to the software infrastructure and architecture and then of course meeting the deliverable requirements in time our timely fashion and the cost or the standard project aspects was the most richest challenge that I've ever seen so this was the best cross-section of the multidisciplinary nature of robotics and I wouldn't necessarily have referred all as robotics at that time. I just knew that I needed to create this thing there was a fundamentally decision-making reasonably pseudo intelligent device combined with all the rich disciplines that I've enjoyed learning about. From that point forward was essentially hooked. I wanted to try and find the space that would give me enough room for growth as we evolved in the technology as I could and that was robotics.
Per: Then you went on to do other things before we get to the modular robotic stuff?
Daniel: Along the way, in that application we actually found that just the single axis degree like single degree of freedom solution was needed. We needed to lift a probe up and down actually in the vertical space and we found a piece of technology that would interweave a handful of sheet metal items that provide, it was a strong rigid device and would give you a nice rigidity. Except this thing would only go up and down and cost anywhere from $10,000 to $12,000 US so it didn't make sense that we're going to pay a single degree of freedom for this and then integrating it with actual sensors so we could get precision feedback on its position and then working our way back through the kinematic chain to understand where the probe was with respect to the radiating antenna systems. It was all too complicated but I've worked on a variety projects from secret comms up to aircraft, worked with Boeing aircraft within the defense sector. I've worked on naval vessels in combat system solutions and stuff like that but I was never really a good advocate for defense to be honest. I was much more interested in doing something that would also provide to the community in kind of way that defense towers but much more on the, I'd like to see technologies that can be commutable and transferable and portable across a variety of disciplines. I loved the idea of the human machine interface. I still love the idea of the human machine interface. The ability to create technologies that can leverage on the disciplines of medical, biotech, robotics and automation and industrial manufacturingand all these different disciplines can now benefit from some platform centerpiece, some basis with which that knowledge can be injected and then installed elsewhere.
Ideally no matter where we put it if we find a commercial strategy for that technology then ideally we can start to disseminate it into each of these industries. We prove one vertical we move to the next. It was really like robotics has the versatility to do that. It’s just at the moment got complexity and it's expensive. That’s hard. You need a lot of time typically. None of these things are available to you unless you have a commercial strategy. I've done a lot of things that kind of fed into this ecosystem but the reality is it's not just that that pushes to want to create new technology. It’s a personal endeavor and a personal grounding with the ecosystem that I was in. I didn't really like being around what was ultimately the public service environment. I didn't really enjoy the not quite ambitious enough tone of the world that I was in. I wanted to do something to fundamentally push the limits of what I believed was possible and my perception of what was possible was significantly greater than apparently then the individuals that I sat next to. A combination of this personal push and this desire to thrive up and do something grander and this desire to see technological challenges benefit our community across a series of disciplines that leverage each other just meant we need a platform, we need a technology that can service that.
Per: How did that lead to you than starting Modbot and getting into modular robot.
Daniel: It starts as a combination effort and this is where the duality was a necessary part of the technology. I met a guy called Adam Ellison who's my co-founder and he is an incredible mind lent to the beautiful mechanical expression creating transmission ideas was something that he did as a pastime in his idle efforts when working for the automotive sector and a variety of other spaces. In reality the birthright of Modbot is this combined coalescing of problem spaces shared between us as well as the personal desires to accomplish something together. Truth be told we started off very small. We thought why don't we create a piece of technology that can test our capability to work together and provide a little servo actuator that would be a significant improvement upon the existing state of servo actuators. When you pull them from the hobby sector they're typically not that high-quality. They can't carry the load you want. They have a regular torque. They have cogging. They have all these other issues and at the same time Adam really wanted to test a handful of technologies that he was thinking. I really wanted to expand the platform.
I wanted to enable an ecosystem that could support the community. This combined effort meant that we kicked off this project about two and a half years ago with a test of a hypothesis around our transmission and an encoder. As I mentioned before the only thing really missing from that is the ability to find a commercial strategy that gives you the time and resources to let that technology flourish because it's one thing to have the idea for something cool. It’s another thing to have people's let's say patience to let you evolve it and let it flourish and let it turn into something real. We ended up applying for the TechCrunch battlefield event. This was done eight to nine weeks out from the actual event itself and all we had at that point in reality was a handful of ideas and some very rudimentary tests of assumptions on the technology. This is then as I say full circle, the technology and the commercial element then came together because we then sprinted for eight weeks to create everything that we needed as far as the vision and alongside this technology piece came software, came the sudden apparent desire to radically simplify the interface for controlling robotic systems, for enabling you to key frame and just move your robotic solution if it's effectively just a precision motion platform from point A to point B.
Giving people an ability to not have to necessarily code but to move virtual objects or to move the physical robot and gesture it and then key frame it. Combine that with the scalability of this solution adding more modules and we've ended up what is this Lego-like module set where if we take the actuator and we combine the fundamental principles of anthropomorphism we think about that same thing of like the human machine interface then we have an actuator is essentially a muscle. We have a bend that represents these joints within the musculature system of our body. Then we have these bones that represent these straight links. This fundamental basis gives us the kinematics of the human body and so what a great place to start. That’s been proven for the evolutionary cycle. We see that we have evolved and we see that this seems to be working for us. In reality most of the systems that we interact with designed to service this form. To bring us forward and now we combine all of this stuff together and we go what we really need is to design high quality fundamental modules that represent these elements of the anthropomorphic form and present them in a package that can have a commercial viability.
Per: By being easy to use, by being easy to reconfigure. That’s exactly why I love modular robotics. The fact that you can build whatever you need from them. You rather than key framing in software and controlling them easily. I know that that's something that the researchers in modular robotics is really working hard on how do we not only make the modules but also control modular robotic systems which is much harder than a general one because if it's easy to build the physical hardware and very hard to program we haven't gotten very far. It has to be easy to do all of it.
Daniel: As far as I’m concerned, what I wanted to see, what we both myself and Adam wanted to see was we wanted to take the rich, complex building of robots out of the expensive research facilities and hand them to the millions. That’s kind of the core of Modbot. We want to make robots massively accessible and for us they need to be simple, they need to be affordable and they need to be useful. To that end I actually realized even recently like a lot of the marketing stuff and jargon that we kind of start to use now is this it's simple, affordable and agile. Agile to us seems to be the best representative term for what modular provides. Modularity gives you see the ability to be flexible, to shift between use cases, to make a right sized fit robot, to design your robot that specifically serves the purpose of the problem space that you're actually addressing. Shift people out of the mindset of worrying about what the robot, like how the robot is going to work and instead focus on the tasks that they want to accomplish instead. These are the high level layers of thinking I guess that come from that.
Per: You now have been working on this project for as you said two and a half, three years or something. Can you tell us a little bit about the journey and then we'll talk about where the project is right now because it's been a journey I know.
Daniel: It's certainly been a journey. It’s funny. I like to pre frame the story though with the reality that every start-up seems to have a shared story but it's all fundamentally individual and unique. We will have to go through it ourselves it seems but the parallels between them are phenomenal but Adam and I met back at University. It was about 10, 11 years ago and we actually parted ways and had our own history through careers and life and so on and so forth until we met again back in Melbourne. In doing so Adam had actually done a lot of entrepreneurial stuff. He had worked through the automotive sector and then done a bunch of start-ups. I had done a handful of start-ups of my own but had worked through large enterprise and defense and then contracted out to be a systems. I’ve done a few as I say companies in the side. As a result we're looking at the idea of starting an entirely new concept at risk and evaluating our life and how that would look if we were to give up everything and give it a go.
TechCrunch was the thing that really pushed us over the edge because we literally presented. We started what we thought was going to be the hardest work stretch sprint of our life. We did seven weeks of sleepless nights to create what we were hoping was going to be the first prototype that we could put on stage at the TechCrunch battlefield beginning of 2014 and at CES, live in front of all of us. It’s a terrifying concept but she was a motivator to the point where we were up until 1:00 AM, 2:00 AM every single night for about seven weeks to try to get this thing across the line right up until flying over to America not knowing what's going on, landing the night before, having about 25 minutes of sleep. When we finally pulled everything across the line branding and visual imagery and basic rudimentary electronics and mechanical architecture and then presented. Four minutes flashed by and they were walking off the stage and we're represented by a colleague actually now who works for us Suriya who handed us a brochure and effectively said, you need to come visit us in in San Francisco because myself and Brady, we're part of an incubator called Highway One.
we went back to Australia after visiting them and a week and a half, two weeks later we got a call from Brady who basically said we love you guys and we need you back here in two months. Needless to say Adam and I actually looked at each other when we read this email actually. I pretty much remember the only expression was well shit. What are we going to do now? In reality what we did was we presented a package that we couldn't say no to and we said to them look if you give us this and if you give us the time we need to sort of figure out how we're going to get there then we'll do it thinking that we were really pushing hard. The only response we got back in the email was sure, see you then to which we went oh shit. They accepted what we put forth. A few weeks later that was it. My friend was living in my house and I was in America. Adam had thankfully convinced his girlfriend that this was all a great idea that we should pick up our life when we moved over. We landed in San Francisco, completely foreign in a new environment and had nowhere to live, no rental history, no credit cards, no access to a cash except through an international medium.
Per: San Francisco was better then but still terrible.
Daniel: Let's put it this way. I was living what I thought was a higher expense than I wanted to living on a two-storey wooden floor, I bought a beautiful pan-European apartment in Chapel Street in Melbourne in South Yarra, beautiful area. I came to San Francisco and I live up a ladder, in a corridor about a single bed wide, about four single beds long for about nine months and paid $1,000 US a month for it in a warehouse. I'd hop over a handful of meth addicts and heroin addicts that lived in the bridge just down the way from us. It was just an unusual paradigm to think that I'd come to the land of innovation, to this city of huge and immense big thinking to have such an incredibly difficult experience living a natural normal life because like we were just challenged in every way. It’s crazy expensive like I was confronted by the size of American meals. I couldn't escape cheese or sugar but no it wasn't that bad.
Per: I heard about bunk beds now in kind of a hostel situation costing $1800 a month.
Daniel: I was also privy to that reality. In fact when I landed in America I had applied from afar because I was concerned about having a place to say when I got here to a communal living facility. It's a four-storey place that you pay about $1250 a month to live in a share room in a bunk bed with about 70 other people that share one kitchen. It was it was terrifying. It was a really awesome kind of collaborative community environment but at the same time it was a challenge.
Per: It was easy to stay late at work when you're coming home to that. For the business it's probably good. Not many distractions at home.
Daniel: The reality for me was with the intense rich amount of brain power that you consume all day you just get tired and fatigued so what you really want to do is just go home and switch off. You want to isolate yourself away from people and allow yourself to be recharge. Sleep, rejuvenate, go back into isolation so that you can recharge the batteries. You don't really get that when you share a bunk bed. That’s when I moved. Adam lived upstairs with his partner and I live downstairs in this warehouse that's in an area between Paccheri Hill and the mission. It was really cool downstairs area. The guy that owned the place had turned it into its own little personal clubhouse but at the back was this little commercial ladder that you climbed up. It was effectively a nook. I call it the mezzanine corridor.
Per: It's like in the head of John Malkovich.
Daniel: It was just like walking through the mini door on floor seven and a half or seven and three quarters. It was funny lifestyle but the reality is it for some reason had no bearing on what we were trying to accomplish. It was the willing sacrifice that we were making to start a brand new idea and actually stomp the pavement and in reality gain the true credibility that is required to know what it takes to work for something and bring it to life. We did this incubator and we were 12 weeks into the incubator. We actually got flown over to China by Highway 1 and we were actually presented or approached by Brady and one of our other colleagues who said you're not going to make it to demo day. We need you to reconsider if you're going to go on stage. We thought what are you talking about? Of course we'll make it. I thought about it a few moments and actually we're in robotics. We have more challenges than everyone else presumably.
Per: You got software on top of the hardware.
Daniel: We have a lot of risk that's probably more the point. What we do in robotics is we stack the challenge. We have a mechanical interface when we create a new transmission solution and we stack that with a custom motor with a custom electronics encoder and board with a solution that doesn't necessarily pass the MCMI and know if we're going to have any communication errors. Then you have to run the software. We wanted to have a Bluetooth configurable and controllable solution from the phone that could run a mobile robot when we got onstage. What we decided to do was we realized well if we do a sprint and we utilize some off-the-shelf technology then we might be able to put ourselves in a position where in five weeks’ time we can deliver what we need on demo day. Adam and I came back from China and we designed and designed and we worked until 4:00 AM every single night thinking that the last time we do the sprint was going to be the hardest time. This was going to be the hardest time. We worked every night for five weeks solid and we found a prototyping shop in Shenzhen that help make all the mechanical systems for us. Everything got delivered in two weeks’ time and then we sat there working every night. I soldered up probably e00 or 400 systems of solder joints that were required for our communication to power infrastructure. We put together the board, put together the electronics, infrastructure and software. Put the mechanicals together. Design the bearing base and the bearing retainer. Putting a system in our form factor and then Adam sat in a dungeon room slowly assembling every single component by hand singularly without an ability to test or check anything.
I created all the electronics and software infrastructure without anything being tested and then on the night of demo day, right up we were doing our first live testing at about 2 o'clock in the morning before we went it seems to be working not knowing which direction the robot was even going to turn in case it was going to run and smash into us. We managed to pull across the line with an LED ridden, we had a dry ice fog and everything on stage and then pulled a phone out of our pocket, stood on stage and Bluetooth connected to this real robot that actually moved and worked. We had 30 minutes of sleep. I remembered getting off the stage and both Adam and I were approached by the CTO actually of the company that runs Incubator. He walked up and said congratulations guys. I honestly don't know how you pulled that off.
Per: Then you said, thank you very much. Where’s my pillow?
Daniel: we fell into a coma. What was funny is we had an investor from LA that had come up that day to come visit us. We were like now we have to entertain our investor. We need to show him all the good things and get him on board with stuff and get him to invest. We have to put on the smiley face. 7 o'clock at night, we had a half a glass of wine and both Adam and I looked at each other and went, I don’t think I can walk. Damn this robot did this to us.
Per: I have done all-nighters and they're magnificent in a way because as you say you can get an enormous things done but after that I mean I'm getting too old for those. I need to be in bed for a week. Poor Adam's girlfriend.
Daniel: I know exactly. She’s in a foreign environment. She’s having to find out if she can't handle visas and deal with finding a job and find a stable environment. Of course that's your rock if you've got a partner that can help you out to get through that. The start-up environment like that it's pretty troublesome because you don't know if you going to get paid. We worked for a year and a half and we didn't earn a single dollar. All the money that we took in investment went directly into the product and rent. That was it. Honestly that was chapter one and I wouldn't change a thing.
Per: This also a life adventure. It’s so valuable to have done that. I've done that in previous projects and that's the best times ever.
Daniel: It gives you perspective, credibility and it actually earns a lot of respect I think. When you can show and demonstrate the sacrifice you're making for something real investors see that. They know that you're committed. They know that you're the right person that aligns with this product, this thing that you believe in so that passion is a requirement I think of longevity, of sustainability.
Per: They also know when that's needed because now you've done two of those sprints. I presume you've done more since then. There’s going to be those sprints in the future too and then there are those sprints for even large companies. I'm sure Apple sweated the first release of the iPhone the way you wouldn't believe. That also shows them that this is somebody that doesn't crack under fire, doesn't crack under pressure. It’s also as you say gives you confidence because I get that once, I did that twice, I did that three times. I could do that.
Daniel: Well it gives you perspective because the more you do it the more you realize there is another end to it. You get to the other side and there's a beautiful, there's a rich fruit orchard bearing fruit.
Per: Should we discuss the product then we'll go to when it's released and how people can get their hands on.
Daniel: That story really is chapter 1 and chapter 2 is now we at least get to distribute our energy now across a team. The company turned into a team of 16 people at the moment based in San Francisco. We’ve just released what we call build one. We’re at the beta technology stage now. We’ve done a first release of the technology we've built, it's beta built. The technology at its core and the bread and butter that we fundamentally offer is a consolidated actuator. It takes a transmission and it combines with a highly customized and focused brushless DC three-phase motor. It has the encoder technology which is our own custom technology integrated onto our own controller board, integrated into an orientation agnostic centrally spun actuator. This is a high torque density actuator and that's its fundamental offering but it's a sophisticated software solution that provides us with the ability to cancel out coving, to have a nice smooth response. We can implement field weakening solutions to increase our speed range. We can have high-speed communications to the systems on the front of our connector with at the moment EK CAT connection. We run a 48 volts DC bus line to it and we can pass the parent coms straight through the system which in itself is make a pretty incredible offering in the hardware because what you've now got is a high torque continuous rotation, low inertia robot joint that can both act as a robot joint but potentially a wheel or something else that you might use a slam or a ground-based movement, stuff like that.
This bread about a building block then combines with different types of bends. We have at the moment the P Bend and an L Bend. The P Bend provides a nice balance point in the system so all the weight distribution is over the baseline. It gives a really good dynamic response but it has pinch points. We wouldn't use this particular bend for collaborative robots. The trend in collaborative robots is real. We know that the stepping stone to a fully autonomous factory is to have humans and robots work together much closer.
Per: I'm definitely of that belief too.
Daniel: The L bend services that. The L bend then introduces a gap between our like let's say bones and elements of the skeleton of our robot systems so that the fingers don't get crushed and it's intrinsically safe. If you then combine that with the software system you have collaborative as an option. To us that's really interesting. Adam and I often talk about that and he's of the opinion that in a collaborative it's a great approach but it is a stepping stone. We should be ready to enable a platform that can allow people to when they want to flick the switch and just make it go fast we can do that. When we need that throughput we can do that. We can provide the feedback to the customers so they know that we're in collaborative mode but it doesn't always have to be there. That’s pretty cool.
Per: I also think that that's absolutely what we're going to do with the collaborative robotics but also duality mode with collaborative and non-collaborative. You could do part of the tasks together and then the robot could quickly provide automation and then we can again do things together.
Daniel: this leads to the other aspects of collaborative are you want to work alongside the human but again if you use that anthropomorphize mechanism the humans more often than not don't run into each other. I usually don't have to run into you to know that you're there and that's because of my rich sensing systems and of course and the mental history that I keep as a track record of where things are in my environment. We want to introduce that into the platform as we move forward. Collaborative robots right now have kind of gotten to the point where we either introduce mechanical compliance like the rethink robotics solutions that we can then have torque sensing and detect when we've impacted on things. Universal robot uses like a lot of sensors and they can have current sensing and these kinds of things they can detect the delta and some basic external torque sensing but ideally it would be great if you don't have to hit anyone to know that they're there.
Per: Eventually that's what we have to do.
Daniel: Vision systems, capacitive systems, ultrasonic systems these will all be part of the mobile platform as we move forward, different modules that you snap into the system that start building the picture of the world around you and this to me is the key to the future of where robotics will start to evolve. Right now we are dramatically improving the intrinsic problem which is the robot itself. We take the complexity away from the user now because we solve the coupling of bearing a transmission and motor and control electronics. We remove all the interfaces which means we reduce the mass of the system. We increase the strength and rigidity of the system because we have the beautiful what we call common mechanical coupling we can maintain stiffness throughout our system but also provide modularity. In the future we need to start dealing with the extrinsic problem which is the world around us, the dynamic changing momentum, the organic sea of stuff that's going on in our environment. That will be this perception system plus machine learning plus other systems that then start to tell the robot what's going on around it. That’s going to be part of the vision as we move forward.
Here’s how I think that that can occur and Adam has actually created an analogy about this in the past. We did a presentation that was put together down in Carlsbad a while back. I deferred to his analogy for this one because it's a good one. Think of the PC revolution. There was a time when the mainframe existed in his first creation and it was a large single purpose, very expensive piece of planned capital that was only really accessible to a variety of people in a server room somewhere. That was computers. Somewhere down the line we found a way to introduce it to many more people. It became accessible and the way that we did that is we actually broke the individual constituents of that solution into modules. We created desktop PCs that had motherboards. Monitors were separated. We had graphics cards and floppy drives. We had each of the individual components serviced by someone that was best to do that part that then coming together but the modules meant that we can now make the computer that you need. It was specific to your purposes either by cost or by performance but either way you could find the you know the little slider line or slider bar for exactly what you need. These individual modular constituents allow you to have this flexible response to your computing needs. Look at what's happened as a future of that. We introduced accessibility which radically increased the technology. It went up. We had more and more solutions being solved. We had more people thinking about computing solutions and it's then evolved into smartphones.
what we actually started to do again is now we go through this modular phase we actually have a consolidation phase again and this consolidation phase starts to look like individual special-purpose but very capable solutions that have earned the benefit of or leveraged on the benefit of this modular phase and gave all of the learnings from this experience to that technology. Now we have smart watches and wearables and gadgets and these kinds of things have trended. That’s really like leveraging on that stepping stone aspect. Robotics is largely in the mainframe era. Right now we have large, expensive, single purpose capital equipment that when it's not being used it just sits there ultimately wasting footprint and they're beautiful and amazing machines but they're not necessarily flexible and they don't allow more minds to be using them to solve more problems. What we needed to do is enable robots to go through this expansion phase. Give it to everyone. Give them a flexibility to start solving problems and we may even find that I think they'll always be a little bit of place for modular but even the Modbot system will start to provide solutions that make sense for our future as well. We start adapting towards deep learning solutions and algorithms. We introduce data players, we provide IoT connectivity that enables you to get sensor feedback on the robot that is sitting on the factory floor because server number four has got a temperature at fault so we report that up to the server and we automatically send out a new server.
Per: Even our integrated devices like smart watches or laptops or cell phones are modular under the hood. I mean hard drives in any laptop they're all looking the same and CPUs, all look the same and all the Bluetooth chip they're all the same. It’s just the packaging that are unique to you and me.
Daniel: That's been the difference about what is the fundamental building platform this is what the user sees, what the customer sees and if you can actually enable the flexible background to start creating awesome solutions that solves a problem then the factory floor in China that makes smartphones can use the modular platform for their purpose but then the giant Siemens facility that has to create large heavy plant equipment like large electromagnetic systems or status for motors for example can use the same tool but to them and from their interface it looks exactly like what they need.
Per: Amazing and we see also the crossover between different areas where one area pushes for the development and then another area benefits from it but that area in itself wouldn't have created it. We see this with the GPU as the need for people to game or the desire for people to game created this enormously powerful GPU systems and then something you can do so many other things with them, video editing or you can do a supercomputing on a regular hard drive or regular desktop machine just because if you can leverage this GPU. We’re going to have these crossovers in robotics too. Somewhere somebody has a strong need for something and then that becomes available for everyone rather than staying in that domain which is what I've done if we had mainframe still where everything is locked in and proprietary.
Daniel: It's actually an interesting point that you then bring up because to me that is then the next evolutionary vision of where this stuff goes because the GPU is an incredibly rich now sophisticated. It learnt from the benefits of this modular tone and has been specialized piece of hardware but where is computing gone? Computing has gone into the cloud and now we actually have a distributed model of processing and computation. We don't longer have to necessarily associate this mechanical system to its processing in a software that have to be co-located. That can be distributed so there's no reason why we can't automatically respond to weather patterns or to some system of affairs or like a volcano eruption that occurs in some place in a planet that automatically gets detected by a deep learning algorithm on the internet that then ramps up automated factory in China for the distribution of emergency aid services. Then they automatically go on to a supply chain logistics company that then drones them to wherever they need to be. Then a robot in itself including the Modbot system could mean you build a modular robot that's sitting on a factory floor in China but you've got a software technician sitting in San Francisco who writes the application code and deploys it over there. When he's ready that data can then be sent to a service technician that actually monitors all of these systems and as I say sends out automatic deployment of each of the modules that need to be repaired to any robot elsewhere. Then if you get stuck you're on the factory floor and you call out Modbot. Modbot can just go well why don't we just remote into that robot and we can help you to build that application. Think about this even more, every time you train that robot you may well be in a position to actually capture the data with which you're training that robot with. There’s no reason why you can't use that data to then automate how you train the next robot because it's learned how you trained.
Per: it's only the difference between what you taught the last robot and what you taught the new robot that you actually have to teach them. With the human you can just inherit that which is what we do in source code and in software all the time of course.
Daniel: It's not so much just the let's say a replication of software like I created that solution and I'm going to copy that software into that robot but rather I'm actually going to copy the way I trained that robot into something else. You actually step a layer above and you go well now I can improve or automate the way that I train the robot. For example if I tell that robot okay, well I need you to drill a hole .it's going to go I've already been taught how to do that. Then when I tell it this is what you need to do. It’ll go oh no, I already know how to do it better than that because I've been taught 15 times before. The last time I was taught how to do this was Germany, that person told me it should be done like this. I now know how to detect a nudge around a hole and I can automatically do a helical spiral and file myself and deposit it down.
Per: Of course also we will all be using state of the art skills because say that you want to teach your robot to do something you can find a person who's best at that thing and have him teach the robot to do it. We’re all going to be running like Usain Bolt because he taught the robot to run. We can all develop and become the Usain Bolt of a certain set of things. We don't have to spread ourselves thin by knowing everything. We can be focusing on what we are good at and what is the core of our business and our needs.
Daniel: That's probably what it comes down to a core. At the end of the day I think we are we are such a globalized and diverse community now around the world and tech is very well heavily connected. I mean there's definitely lots of wells that aren't still in the world connected up the way that we know we can in the high-tech world but do what we know how to do best and let's leave the rest of someone else but more importantly let's give them the tool that enables them to do that. I think robotics has been something that's been as I said before stuck in the really expensive research facility. If we can suddenly give a lower cost simpler technology to many more people what we actually do is enable an entirely new class of roboticists. We now have different types of people coming up with the solutions in new ways and new perspectives for new challenges and new problems that we haven't thought about before. That’s kind of why we went for this. We believe that the platform is a necessary stepping stone to help people to do that. One thing I'll add into the platform part of the conversation that I think is important is I don't believe any platform is truly succeeding nowadays unless it is open, unless it is accepting of the fact that it does not exist purely to keep you in the platform. The goal really needs to be that you provide a tool base, a certain sort of platform layer in such a way that it is accessible to you as a user and that is accessible to every other product that that user might want to interface with.
Per: I'm definitely with that being open and that's coming from somebody that grew up with Microsoft and Apple. I just hate the closeness of these systems and being open is just essential.
Daniel: I mean open has this twofold effect. It provides a much greater reach of accessibility to be honest because now we don't have to solve all of those individual problems. As you say before we let the people best to do that do that like ripper manufacturers or perception system makers or different deep learning online cloud-based system providers, all this kind of stuff. We focus on where we're very good at and provide the interfaces that then leverage the benefits of each of those technologies within our platform but furthermore because it's open you don't necessarily have to use our platform you can still use our hardware but run it with your own controllers or you could use that controller use our software but on other people's hardware. It incentivizes us as a company to have to keep innovating so we what we're going to do is be motivated to prove to you that we keep making cool things and then it's worth you being invested in our platform because our platform get better, makes your staff cooler and we'll never lock you in and we have the open-armed personality and culture of a company that makes it look like hey, this is a great place.
Per: That also prevents you from becoming complacent. That’s probably the worst thing that can happen to a company. If you know that that pressure is there all the time you respond to that. If that pressure wasn't you run the risk of becoming complacent and saying well, they have to use my system. I’ve locked them in. The user always wants to be free in the end. They’re always going to break out and do something different in the end anyway. Why try to lock somebody in when you really can't and why get that badwill of trying to lock people in. It's just the disaster.
Daniel: It's human behavior I guess at the end of the day. If I speak for myself as an individual I know that if I'm given more option and openness I'm more comfortable and I'm actually more willing to hang around.
Per: I'll stay voluntarily much longer than if you try to force me.
Daniel: To some extent I always enjoy like Google when they provided there data liberation service. It was a project and a tool that we just want to make sure that you can do whatever you want. If you want to leave he's how.
Per: I used that just for backups and I think that was great. Knowing that that's there made me feel better about staying. Knowing that there's a door where I can leave anytime I want to makes me safe.
Daniel: It means that we're still willing to let you be in control of your fate and your situation. We have to make robots like that. Robots is a funny term and concept in general because what is a robot? To me it's something that I think is going to always be an evolving concept, an evolving term but we are heavily dissuaded by Hollywood because what we see in movies and cartoons and the characters and the author's creations and depictions of them seem to always have this anthropomorphised version. For some reason I think humanoid robots are going to exist someday even though it's quite debatable that there's no real economic viability or feasibility for them to exist on a factory floor because they don't necessarily operate better when they're anthropomorphised. They’re better off having wheels to be honest or something like that but we seem to want them to exist so they will.
Per: Well I think that's also I mean human beings are obsessed with human beings. Monkey want to see what monkey do. We love to see the human form because our brain is predestined to look for faces, look for the human form because humans have been such an enormous part of our history. I think that probably has more to do with that than with anything else.
Daniel: I would agree.
Per: It's also very well-known form if you make a robot looking like a human you kind of know what that robot should be capable of. It will take a very long time before it's capable of that. Before you have a gymnastic robot that can beat gymnastic in the Olympics that's going to take millennia, decades if not millennia. We still think it does. If you see a robot that looks decently like a human being you presume that it's going to have the capabilities of a human being even if it's nowhere close.
Daniel: Which is funny and this is the uncanny valley, if you get too close and you're not quite there and something looks that good, I remember reading an article about Pixar and I actually found that when they created their animations to make them too real didn't serve any benefit so they actually purposely make them animated, make them cartoon, make them naturally non-threatening.
Per: Uncanny valley is a big thing and I'm of the persuasion that of the old I have the idea that we're so good at seeing these small abnormalities that it's going to be next to impossible to create the robot that could physically fool another human being in a room sitting watching them saying is this a robot or is this a human being. I think we're so tuned to the smallest detail probably from the dark, this probably comes from the fact that historically we probably all surprised the people that were sick because we didn't have modern medicine so when somebody started behaving oddly we shun them out of the group.
Daniel: We avoid because I guess that's the safest way to stay safe.
Per: It's like when dogs do this they never show you they're in pain until they're in very severe pain because they would be shun out of the pack.
Daniel: I can combine to your concepts here or rather they combine in my head where you talk about when people are sick we tend to shy away from them.
Per: Or historically we did.
Daniel: We did but we're also obsessed with humans so what I find interesting is when you see a Maine dog or you see a cute little puppy that has had a swollen up eye and it's had this disastrous thing occur in its life at some point for some reason we find them cuter. We find the poor little guy needs to be helped.
Per: Like the scar on the face of Harrison Ford.
Daniel: Those kinds of imperfections they give us reality but when we see things that are a little bit unsettling in the human form it bothers us. When we see someone for example that has a dismemberment that concerns us. A humanoid that has a dismemberment as well for some reason also bothers us. If we see a T-1000 standing there but it's just missing this much of its arm we don't like the visual image of it but if I saw a three legged dog I don't feel in the same sense of offense at all which I shouldn't really feel any way for humans.
Per: I wonder if we're projecting the pain of losing our own arm or we are seeing that this T-1000 is losing an arm. Oh, I don't want to lose my arm because that would be terribly painful. Robotics is teaching us so much about ourselves and so much about other areas that it's just amazing.
Daniel: But to me this is the reason why what I anticipate is the future of the let's say robot personality is likely to not be so much the humanoid as much as the anthropomorphization of certain characters inside of these robots.