The Future of Advanced Robotics feat. NVIDIA, SIEMENS and TERADYNE
we have some people here who are working very much with a future forward-looking angle I'm going to let all of you introduce yourself a little bit in a moment um deepo would you like to say just a couple of words about yourself and the job you do sure uh first of all thank you for having me on behalf of NVIDIA what a wonderful building and what a wonderful City uh my name is De uh and I lead Robotics and Edge Computing at NVIDIA I've been working at uh NVIDIA had a great Fortune for the last 11 and a half years or so and uh we working on the latest and greatest Technologies using artificial intelligence so looking forward to chatting with you R yeah also thanks for having me here um I'm for at Sean responsible for the automation business and I just want to reflect to what Kim said you know my passion is really as market leader for automation to bring automation to the next level to really fulfill the major topics we have in society and therefore we need to bring it technology AI technology but also a lot of activities around tery around around robotics into one system and give that to our customer to Industry at their hand to make it easy to use and that's my passion um ell Kumar I'm group president for tadine Robotics and I'm so excited to have uh great Innovative companies like you are and me in on our platform uh I I've been here slightly less than uh J uh started in July of last year and most of my career has been with larger companies and I'm hoping to bring in that unique Insight of how do we take this great platform we have to solve problems for the large customers the same way we have sold for small and medium customers for a long long time thank you we are going to go deep in a minute but before we do that and especially for our podcast re automated which we will use come back to this discussion for could each of you tell me very shortly one thing that really excites you about robotics right now Deo wow um I think there's two simultaneous forces that are really happening right now that's going to change the trajectory of AI and Robotics I mean AI is not that new robotics is not that new however the Advent of generative AI in the last 12 months coupled with the simulation and digital twin technology both of those things are really at a Tipping Point right now and combination of that I think for the next five years is going to change the trajectory of Robotics Rina well a lot of the unsolved automation tasks which can be solved using AI using automation you need robots and therefore you know that combination that fascinated me how we bring those things together to solve the last remaining task to be automated um what excites me the most about the space is uh the collaboration is at the core of how we solve our customers problem no one company has all the Technologies needed to solve this problem and we'll need to work together to uh understand the problems and solve it at a very fast paase and we have talked about this building is being built for collaboration after this we're going to go and look at some robots I hope all of you will go and look at some robots you're also going to see some real collaboration going on with our robots and our partners appliations and our robots working together but we also work and collaborate through real company collaboration and I'd like to think a bit about what collaboration means to you and NVIDIA is an interesting one because people don't naturally think about robotics straight away but I know it's something that you're really working on now what does collaboration mean to you in that space yeah I mean uh NVIDIA we are a platform Computing companies we you know we in the area of robotics for example we don't build robots we work with every company in the world that's building robots or robot platforms so what we are bringing to the table is number one you know people think of NVIDIA as a chip company or making hardware for all sorts of computers in the data center and servers and also even uh Edge Computers we definitely do that and in fact for robotics uh you know the robot brain we call it uh NVIDIA Jetson most of the robots today you know use that but also for the training and simulation of these uh robots for creating the AI for simulating this AI in the virtual world digital twin uh so we provide that infrastructure if you will but more importantly than that we provide all the acceleration libraries all the AI Core AI capability that gets integrated into the hundreds and thousands of companies who are building robotic platforms and robots and so our go to market our complete approach is it's 100% collaboration because we are a platform company which is great for terod Robotics and we've also been collaborating very recently with you a lot and uh Rana we've also been collaborating with Seamans so super interested in the way that Seaman approach is collaboration between companies well first of all thanks for Universe Roberts being a very trustful supplier to our factories because we are we are a customer of universe Rober and I was very pleased to hand over I think two years ago the supplier award uh to Universal robots so we are leveraging that and we want to build worldclass factories by ourselves and world-class factories need worldclass robots and that's the reason why we are very mainly using using Ur um but not only in the factories but also when we want to demonstrate where technology moves we're going to also mainly using URS in our demonstration kits and so on so that's also good and we collaborating there we might collaborate in the US of getting closer showing together the C customers if you bring Automation and Robotics nii together what can be done making that tangible making that real and one topic where where we believe our role is we also don't build robots and we always have been asked why you not building robots everybody else builds robots do automation robots I think there are so great robot companies on the on the market why should we also build robots I want to want to integrate and provide it a seamless integration of robot in the complete automation solution and that was the reason uh we are collaborating with with uh Universal robots of bringing a standard command robot interface into the market and why is that great because we currently have in somebody programming Automation in in the in the seens world we have 200,000 active user every every week programming and they are capable to program a automation controller not everybody is capable to program a robot but you need to program a robot you need to integrate the robot into for example line structure which you have in factories so now we we collaborate and say can we you know access the robot from the PLC program seamlessly and we just brought that together out of the the market I think for that and I really looking forward of you know making robotic much more accessible to a broad audience of all automation users and it's real theme I think about saying we can't do everything in fact we're building one type of robot primary at Universal robots and we're not trying to build other types of robot and that's where we go to for partners and udw when you look at the ecosystems that really generate value to customers and partners in the companies that you're running where do you see the ecosystems going and in fact we have a lot of ecosystem Partners here they'll be interested what's the future for the ecosystem no great uh I I think on the ecosystem side there are two ways in which I see us uh moving from from where we are today like we have one of the largest ecosystem in the robotics World by a huge margin number one is what Rainer just touched upon this is bringing in um bigger players who can help us bring in the the um like open up our ecosystem to new uh customer problems like with this srci which I understand we have been dragging our feet for 5 years we finally got through that the these are ways in which now will bring in bigger players like seens like NVIDIA we have several Consulting companies which we are Daniel is working hard to uh collaborate with these are ways in which we will get into new set of customer problems for large companies and medium Smalls siiz companies which we have not had access to and the second one is for the ecosystem Partners we have we want to bring in AI as a huge new tool to solve problems which are of higher mix even unstructured environment and that is where um we are working closely with NVIDIA to see how do we make that whole process more standard like give them a kit we uh started uh Eric is here um a fund where we will uh pick some some of these small siiz players who can also bring in these AI Technologies quickly into our ecosystem and I think that's a great bridge because anyone who knows these companies will be sitting here thinking when will they get onto the topic of AI so let's do that um before we get really OV excited about AI I want to ask you deep is there anything that you hear about AI on the radio TV that just really frustrates you well say exactly frustrating um I mean deploying AI in the operation technology world can be a little bit challenging and this you have to be a little patient but you also have to work on the most important problems and go deep and that's really what we have here the collaboration with seens and terod we've been working on this methodically for several years now so when people think of robots typically it's been Hardware in the past and it's about mechatronics in the past right so they think of the robot computer they think of all the Mechanicals and the electro Mechanicals right electricals with AI what's happening is you want to make sure that AI is the one that's helping you solve problems that was not solved before right but to do that you need to go deep into Problem by problem there isn't one AI yet maybe someday in the future but not yet today that generalizes across every particular task so what we do for example you know in the case of of U you know Universal robotics of course it's all about manipulation right so there's many problems in manipulation but one of the first problems with manipulation is what's the best optimal way to plan the trajectory and and you know to plan you know and what's the best way to move around obstacles and it sounds simple for us humans because we know exactly how to move an arm or go around it right but it's not that easy for uh for a robot to do all of that so we've been working with uh you know the uh Ur team how do we Infuse AI into Universal robotics software technology which is what's used by every compan is using Ur right how do we embed into that that's one example in the case of autonomous mobile robots that are moving things from point A to point B to point C it's all about operating in unstructured environments and navigating autonomously right and it's and for again for humans uh I mean we have our eyes and we have our sensus and we somehow know how to go but it's not that easy for a for a Rob robot so you have to build these uh you know AI capability for autonomous navigation and then with seens you know seens has been you know automating automation for should I say a century plus a century no no tens of years at least tens of years tens of years right okay uh several decades now right and and and the way seens does is they have you know amazing hardware and wonderful tools as well so so how do you infuse those tools with this AI technology and these things uh take time and it's not a frustration it's just you got a patience to to get all of these integrated so that as you know we have a platform we deliver it to another platform company which goes into another platform eventually it goes into robots so so that's just you got to be a little patient but but go and solve problems the most important problems in a deep matter got it would well what excites you the most about Ai and Robotics I think it's what you often call physical AI sure so yeah I I am going to use this term physical AI as opposed to digital AI because here you're taking AI to a whole different Fidelity actually doing something um different in the physical world there are three ways in which we see helping our customers uh number one it is adding new capabilities to our robots number two it's making our robots smarter and U like smarter path planning or navigation pallet detection there are so many uh different ways in which we are just solving lot more problems in unstructured environments and number three is it is making um our collaborative robots which are known for safety it is taking our safety and reliability to a whole different level and Sean rer we've been collaborating on AI based technology which is super exciting I just want to come back back to what Deepu was saying about time when you start thinking about time in this for me as a desk worker AI started in 20123 and it just shot shot shot off and that's not the case for you and and Robotics and for my engineering colleagues here but where are we in that timeline are we well into it or are we just getting started well I think there are two ways of on AI um the one AI which really starting with the one which is starting to scale and you have a lot of anecdotes you have a lot of proof of Concepts and you can show here and here and here I think what we need to look at how we scale it and how you make it industrialized so industrializing AI where you can deploy a model you can inference the model you can monitor the model and that in a very very automated way and we need to do that in order to scale we need to do it give tools to the hand of our customers where we don't need to be in a expert in order to do AI so it need to be very easy maybe without AI expertise you can operate a system which is based on AI this is where we need to work on to get to scale because else is always a niche where experts are working on something and um that's the reason for example we we we show also Here With Ur where we um want to enhance the skill of automation like you said you know a human knows how to grasp something and put it somewhere um and now you can train the the the AI the object how to grasp it we didn't want to train the object we want to train the skill because my view is I want to enhance automation to the next level that means automating the unpredictable automating a lot size one because currently what you automate is mass production if a you have a conveyor belt and that you automate and produce the same thing over and over again perfect for automation automation is not good if you don't know what to automate and you might need to do that because markets are so Dynamic you need to produce today a tomorrow B and you don't have time to reprogram and you want to automate maybe a lot size one if it goes into individualized medicine if it maybe in the future goes into repair if you want to repair something it's a lot size one because you cannot pre-think of what is defect so you need to bring automation Next Level and there you need AI because the robot on the machine itself need to detect what to do in order to fulfill a task and the topic of grasping was a skill we trained the skill of grasping not the object the skill and with that you want to enhance automation to more skill-based automation system in the future that can use robotic or cannot use robotic but we want to move it to that level because at the end I want to have not a rule-based automation I tell the automation you need to do this this this if this this problem comes because then I need to know the problem I want to give a goal based automation do this how you do it you figure out by yourself that's the one area where where I'm very excited to move automation forward now when you AI came to your desktop this was very much about generative Ai and uh generative AI is another very very interesting field which probably is a catalyst also to Bringing AI um to the OT world because suddenly you don't need to be an expert you can change a lot of things by Ai and um we are very much into that and we just showed on the on the Anova Fair um that we have a co-pilot V can program in the future controllers if you now have the standard robot Library With Ur you basically don't even need a PLC programmer to program it a I going to write the program and you only test and verifying it and that probably will accelerate uh the usage of AI significantly also then in the operation technology I think one thing that's really interesting when we get onto AI is we really swiftly move into machines and how amazing machines are our technology which is good it's what we're here for but actually if you look at the vision of our company we talk about creating a world where people work with robots where people work with robots so ell what does that mean to you what does AI do to the human interaction with a robot and with their jobs so um I I see AI helping in multiple ways with human interaction number one this collaborative robots took the safety to a whole different level number two with this uh now natural language speech recognition now those uh you'll be interacting with robots very differently third one is this emotions detection which is a fast evolving space where now you'll be using these robots uh inhouse for medical care they'll be close to you not in this formal work environment it'll be when you are in pain in a hospital or you need some some kind of care or help at home that uh AI will bring in those emotions detection to take these human machine interaction to a whole different level super so we thought about the the future a little bit and the future's arrived and we know that but let's just bring it back to the real world and what you see in your job so which you've been in your job for how many months in this yeah 10 months 10 months and you came from factories before so you have a long history of manufacturing experience when you go out now and you're traveling and you're seeing customers and partners what difference are you seeing our technology making out there like I'll I'll first start with what rer just uh uh shared and we had not talked about this but it is something which large manufacturers we see as the new reality that lot size of one the fact that we have so much variability that economies of scale which I lived with in my whole career from General Motors to G to Honeywell and you are seeing that in you had that in seens it's gone this is the new reality that lot SI of one the fact that U the problem for for large manufacturers is the same as what small and medium manufacturers they have lived with throughout their life which is so much variability you give me these industrial robot which can do the same thing in a very high volume way I don't have those applications I need to tweak my design continuously shelf life of those designs are a lot shorter it it what really excites me when I go into uh the world and talking to customers the collaborative robots now combined with AI has the solution for this pain small and medium customers lived with and the large customers now it is a new pain for them and you mention that you're already using cobots in your factory and we know you have some fantastic we have great pictures of uh Seaman factories with lines of Cobalts in them can you say anything about your experience as a customer in terms of implementing and bringing in that kind of Technology into your own factories well the the last picture I remember with a UR robot was at the Hova Messa where the German counselor and the Norwegian president was was on stage and we used a we said we make it very fancy and we used um AI speed recognition to move the robot faster the UR robot faster and it didn't work out at the beginning um it was a bit embarrassing but it was because the end you only should have said hello Ai and then it would have started like you say Alexa and then it starts and it wasn't activated so it didn't work but that was the last time we had Ur and seens on the picture and but yes you're right we are heavily using a um Mur um in our factories and I think one showcase or Lighthouse is our Factory in alangan where we have fully integrated line where we really basically could switch off the lights and we're be using drives and inverters really from from the end to the from the beginning of of the of the all the assembly to the the testing the packaging putting it into box so everything fully integrated um and I think what is also what we try to do here is I think you are is always very strong having that individual task to solve in a small midsize company bringing you are into a complete line connecting everything putting a complete safety concept into it simulating the complete line up front in a virtual world and then you know deploying it into the line or further optimizing the line I think that we can uh very well show in in our facility in in alangan and on top of that Sean was last year announcing we're going to invest half a billion in that site and we want to re the first side of really having everything and we call it now industrial metaverse where we really want to have a complete simulation you have all the data accessible we already accessing now the data of You Are of the different robots because at the end you want to have a digital twin of the entire Factory which you're not only using for design purposes to optimize before you want to use that in real life so you want to get the data out of your system of everything have it in the virtual world have it in industrial metaverse and then further optimizing or changing things or analyzing thing op and so on so that's the next step and uh we looking forward to do that also very closely with NVIDIA and With UR let's have another question quick fire um key drivers for these Technologies so putting aside the the robot business part if you are in a factory today what's your key driver for automation I think uh there two things uh generative Ai and digital twin and and I think you mentioned it 2023 was the year for you but it changed and that's really because of chat GPT for the first time ever you don't need programmers you could just chat with you know text or voice so that technology is coming into the operation OT world if you will right and and generative AI is going to change robotics Automation in two main ways one is the user interface you don't need uh robot programmers anymore you can talk to a robot or you can talk to the automation or you can command in terms of you know texting it so that's on the UI side and then the generative AI is more accurate and it's also turning out to be more generalizable meaning you don't have to train on one particular task just like Chad GPT you can ask it about anything you can ask about the weather you can ask about your family you can ask about something right anything it's generalizable fairly well so generative AI is one of those you know driving forces and then secondly I think we talked about you know we're working with seens very closely and now we're also working with uh you know um Universal robotics um simulation is actually a key tool it's not a new thing you know simulation has existed for a long time and it's because it's faster cheaper safer to do in simulation versus the real world but now with you know this industrial metaverse as rer said we have this platform at NVIDIA called Omniverse which allows you to train AI which allows you to get the right model the right physics and you the right you know manipulation and also the right visual Fidelity that you can try thousands of simulations in parallel roughly free compared to in the physical world so so that is also going to transform pretty much all the automation Ryan a key driver for automation that you see in factories I just me to reflect to Kim I mean it it's companies want to produce very close to the consumption because that's from a sustainability perspective a a major aspect so you need to produce where the people sit and that could be in high cost countries so the problem now is first of all it's high cost and you want to produce to a competitive price and secondly we living in an aging society and you don't find the people anymore to produce so if you really want to move forward into sustainable production keeping industry within um kind of developed countries it's very very important to get Automation and in order to get there AI is now really probably a game Cher where you can solve solve a lot of the problems you had you didn't have the people who want to work in the factory you don't have the people to program the automation um and you don't have the flexibility to automate what you could automate in the future having that lot size one aspect in place and that's basically what drives our customers uh forward and the last topic maybe that's also very important to mention the aspect of sustainability so um and and then the question is also how AI going to help in the aspect of sustainability and I just was mentioning my dream would be really maybe before I retire we have automatic repair so we have robots who identify because if I I take my washing machine and uh you know I call somebody then maybe somebody's coming said ah you know you better buy a new one repairing it doesn't make sense and don't find people to repair it uh if you think about um what we are working on really as we speak um car better battery for for EVS you want to not throw away the entire battery uh pack you want to probably maybe take a cell out repair it have a second life um but you do that you need to have electrician because this is high voltage so you have a first problem you get a battery which might be defect what do you do now you find electrician who discharging it can you do that automatically then can you disassemble it can you do that automatically can you have ai identifying what is a problem in that battery maybe using digital twin and so going more into automated repair in the future um that would be also something a big big lever for um for sustainability repair before recycling great we we thought very hard about how to take questions today whether we were going to have people raising their hand in the audience uh but what we did is we turned to our social media followers and a few days ago we said we got this panel coming up did they have a pressing question so we've chosen a couple of those questions to fire at you today um udwell safety how much Focus will we see on safety when it comes to coroporation between humans and robots so the whole field of collaborative robots was created with safety as the number one focus and I feel this will be one of the key differentiator for this new automation technology as we bring in Ai and other ways to get more value out of Robotics okay uh and and I'm going to take one because people need to go and have water and refreshments but um deepo this one I'm going to send to you so somebody wrote a common issue I've seen over the years at various facilities is poor planning with production Logistics machine placement shipping will AI be able to optimize efficiency in all aspects of production I think the answer is absolutely yes and this is where the digital twin comes in and you know you see the industrial met was that Raina was talking about you can actually plan the layout in simulation and operate it and if it doesn't pan out the way you want you change it in simulation before you decide to put it in the physical real world so I think that kind of work is happening as we speak right now I'd like to thank our panelists for coming and I'm going to just ask Kim as well um can we get Kim's microphone on as as well Kim you've been listening and I can see that you've been taking some notes at the front is there anything that strikes you from that conversation yeah I took some notes because you asked me to take some notes you see how it works here so yeah I um so thanks for sharing really interesting I think and I I got a quick summary of what I heard and the takeaways I heard as well for us and the two Key summary points is AI combined with digital twin simulation combined with robotics well over the next 5 10 years probably change everything as we know it today it's pretty exciting if you're especially if you're at the Forefront of this which we are the second thing is in order to achieve all that collaboration between Cutting Edge tech companies is absolutely necessary like what you heard is that NVIDIA is not making robots see seans are not making robots we're making robots but we're not doing what seens and NVIDIA does so we stick to what we do best and then we get together and we combine it into something magic and what we need to do is to make through scale make robots increasingly available and accessible to everybody and we need to innovate to solve problems and unstructured um environments which is one of the big challenges in automation today and we need to take all that learnings back into the actual platforms that we individually do so anybody can get an opportunity to use that our partners so forth so the takeaways I took is the opportunity I think out there for all of us is nearly endless we're just starting this whole journey in fact we hardly started if I'm listening to what we could actually go do and two is we have the tools we have the skills we have the talent as well so I think the last thing we need to do is all of us collectively we need to step up we need to work together and then we go solve these problems head on good great job with the notes that's good good notes thank you to our panel thank you very much
2024-05-23 21:51
