Automation Education

Please welcome to the stage Intrinsic's Wendy Tan White Brian Gerke And our moderator today is Haya Jan Camps please welcome them to the stage [Music] Hello everybody thank you for joining us So Um the world of Everything is getting more and more Automated right no Code and low code is Everywhere and yet if you try to program A robot Things get very tricky very quickly I'm Joined by two of the smartest people I Know in robotics to talk about what is Happening there when you can tell me a Little bit about what you're working on Yeah so the aim of intrinsic is actually Democratize the access to robotics so as You said the at the moment we do a lot Of digitization to actually make tasks Happen how do you actually make that Happen in the physical world and the way We're actually starting to see that in Digital world particularly with the Advent of AI shifting when AI really Becomes embodied fully how is it Actually going to touch our physical Space and for with intrinsic we're Actually want to let you know a normal Developer or a process engineer actually Build a robotics application very Quickly and one that can really make a Difference for you and your real world

Task you're actually trying to do Yeah and what is the contact surface There between what you're trying to do And what's happening out in the world The we we've actually enabled a system Such that we you can have a software Platform to build an application in Simulation that can actually be deployed The same application that you see Running in simulation will actually run Physically onto deployed cell work cell As well so you can actually do it you Could literally use a web browser to Design that and actually be running a Work cell in you know across the other Side of the world if you actually chose To Um so we're really opening up the access And who can actually use this and design It and the applications people are using Could be anything from manufacturing Assembly Logistics Um we're seeing companies in our recent Beta program where we started we're Seeing companies doing things in lab Automation Process Engineering we're seeing them doing it In agriculture small businesses and Large so it's quite it's been quite a Powerful um tooling and what other kind Of robots like what like am I Imagining The type of robots that you're actually Programming this way

Um we're both looking at industrial so Industrial grade robotics so you know Really quite large physical robots for Example like Camaro robots right through To smaller cobods like Universal type Robots so it's actually we're actually Trying to make it agnostic both the size Um and the type of robot we're actually Able to Um use so software to Hardware Agnosticism is like whenever I talk to People about that you get this that's Almost impossible Brian are you trying To do the impossible Uh maybe but uh I think I've got a I Think we've got a really Um great track record to build on um so The the team that I was leading before We joined intrinsic it was is open Robotics where we developed uh what's Called Ross the Robot Operating System So it's perhaps the best example in the World of a software platform that is in Fact Hardware agnostic in the sense that You can take you know different robots Can look different but if you break them Down into pieces they have common uh Kind of equivalence classes of Functionality so you've got you know You've got cameras you've got lasers You've got Motors you've got encoders You've got four historic sensors you can Kind of put those things into buckets And then have interfaces to them and

Then as long as you define those Interfaces well you can start to write Applications on top I mean this is what Has enabled uh the personal Computing Industry it's what's enabled mobile Devices these things if if you look at The details they can be different but You can have this abstraction that Allows you to write these applications That can run on even different physical Embodiments so you can even you can you Could write an algorithm that makes a Robot move around the world and it could Run on a wheeled robot or it could run On a legged robot as long as you've got The right details implemented underneath So I think we've like we've actually Figured out as a field how to define Those abstractions pretty well over the Last 20 years and what we're doing at Intrinsic is really taking that to the Next level and like Wendy was saying Making it really accessible to everyone Taking those ideas and presenting them To you in a way that uh just through a Web browser you can access those Underlying abstractions and write these Applications that can run on many Different robots And so am I imagining this as different Robots working together on a on a common Task or you're using these tools to Program each individual robots and They're just going to have to fight it

Out among themselves uh so I left Because I I did a whole dissertation 20 Plus years ago on multi-robot Coordination Um and I'd say as a field we're really Still not there uh in in the sense of Having robots that are intensely like Really cooperating with each other as Opposed to just working kind of near Each other uh there there are still some Deep research topics there that in fact Some folks on our team are also looking Into I'd say in the near term the There's so much value to be had from Just having a robot do a task and do it Really well that we tend to focus there We will I I still hope that we'll get to The point where we'll have you know Teams of robots really working Effectively together that's going to Take a little bit of time And what is the Programming piece that like it's been a Hot minute since I've programmed the Robot it was very very hard back when I Was doing any sort of automation stuff Like what is the the pieces that you are Adding to uh this mix to make that Easier I think maybe the analogy is much like Um my first company was actually about Was it was a SAS based website builder For his first in the world and I pre-vet You used to if you want to build a

Website you'd have to build your own Server your own Middle where add your Own front end add your own Commerce add Your own we're actually creating a place Now for intrinsic which is in a One-Stop Shop you can actually have all of that Underlying Stack there and you can basically use a Either a graphical effectively a Graphical user interface to build the Workflow you want for the application or If you want to actually go in and do a Bit more of a code level we also provide A better user interface for actually Even coding and scripting at that level And we also make it easy if you do Things like AI skills and components as Well as different types of Hardware Um so if you want to add something in For example like computer vision also in A more sophisticated way or force Feedback skills where you can feel your Way into a task and that's also going to Become available to you and we're Opening that up so other skills Developers can actually add those skills In too and different types of hardware So we really want to actually help the Ecosystem as much as the end user be Part of this and to what degree are you Able to do that across all robots I mean This I keep being hung up on this harder Hardware agnosticism thing because you Know that feels incredibly hard I think

Much like other devices you know it Still requires some sort of integration Driver to add to a platform to do that But I think we've abstracted it enough It becomes more of a production Engineering job now to add that in Quickly rather than have to customize Something in a very sophisticated way And where we can we will use common Interface to do that for example with Brian coming on from over robotics Ross Has some of those sort of apis exist Already that abstracts quite a lot of Different types of hardware and of Course now we're together we've actually Taken advantage of that too I will Continue to improve it yeah I'll say That the I think the probably the view You have of the you know write-in Automation software for like industrial Robots collaborative robots kind of the Old I would say the older way of doing That is to develop it all really in These uh company like kind of Proprietary specific tools and even Sometimes specific languages that were Developed by the manufacturers of those Robots and then the problem is you then You can't really Port that to another Robot um but in point of fact if you Stand back and you look at the robots There they might be different colors They might be made by different Manufacturers they the the links are

Different links and so on but really at Like uh at the level that you need to Reason about that robot in order to Control it to do something useful They're not that different and so we can Write these abstractions that sit on top And let you port applications across Them And so to what degree are you able to Take one robot out and swap it in with a Different one and just say well yes it's Slightly different but you know it can Lift the same amount it still fits the Purpose let's go we've actually done That with some of our initial uh Application examples that we've already Built so we can uh you know we can take In like Wendy was mentioning we support The you know Universal robots the Collaborative robots we can we can swap That robot out for say a robot from a Manufacturer like Kuka and you know They're going to have they're physically Different but we can get the same work Done uh between the two robots and this This is actually a potentially a game Changer for people who want to use Robots because you really want to have Free choice right it could be that next Year for whatever reason you want to go With a different vendor for a new robot Line and you'd like to be able to reuse The work that you put in to develop the First one and that's actually quite

Difficult right now Yeah no that makes a lot of sense online Prep call you uh mentioned something Along the lines that that software Really is the frontier of Robotics right Now does that mean that Hardware is Pretty much sold like do we have all the Hardware robotics we need No I mean You know the physical world is still Mechanics involved there's physics Involved yeah there's real world forces Involved Um and those things still need to be Sold so I think one of the one of the Things we are actually working pretty Hard on with in partnership with other People with folk like deepmind for Example uh with is in the area of Simulation as well and actually how do You get more of a real world uh modeling And simulation than um because it has to Be accurate it has to be accurate you Know right if you had a robot eventually In your home If it's a couple of degrees off it's Going to hurt you and uh that that isn't What we need so we're still having to Move forward in that I think with the Advent of compute the step changes in Compute right now yeah and actually to Be honest Ai and modeling there too I Think that's going to improve very Quickly now

Um so we're also working with folk in The arena well and this is the Interesting thing right so physics are Pesky humans are soft and Squishy and Robots usually aren't and Um like in industrial like in building Cars you see all those big robots all The time but the sense I'm getting is That part of what you're trying to Accomplish is bringing robots into Smaller uh processes and products and Like where it's either built by hand or Like semi-automated but actually getting More automation there is that part of The goal that you're trying to Accomplish I think very much so because we're Looking to democratize the access and Democratization means it's not just to Preserve a big you know let's say Automation Automotive factories but Actually can we how do we bring it to The small businesses to the small Warehouses uh you know to the makers to The uh Retail to Commerce Etc and to do that you do need to Different scales but different different Accessibility Um you want to be able to almost being Bring Ace how what would happen if you Brought a small cell into your home how Would that work and actually we're Pleased that you know 60 of our beta

Users are small businesses not just the Big companies we also do have you know The the automotive and the electronics Etc Aviation but we also have six Percent that are actually smaller Businesses and actually in that we think We'll change the industry and that There'll be more players who can Actually build the applications for Those businesses the smaller end user Businesses but I'll say that we can't we Can't do it by ourselves I mean you just Had uh Gil Pratt from Toyota Research Institute and mark rayberg from Boston Dynamics AI Institute up here and the Work that they're doing I think you saw Some of the examples of kind of Next Generation Dynamic systems uh from Mark And also Gill's team is working really Hard on how to do some of the tasks that Like Wendy was just alluding to where You imagine having a robot in your home Doing something for you to help you live There longer and being able to do those Kinds of things safely there's a whole Bunch of work that needs to happen there Actually back in the hardware so we need The hardware to be intrinsically safe we Need to understand how the control Algorithms are implemented we need to Understand what kind of motion planning Is required so that like the way that I Think about it is we've got a whole Community fortunately who's working on

Some of those underlying pieces and as They become available we'll we'll be Collaborating with them and we'll be Able to expose them to a really broad Audience through what we're building at Intrinsic yeah you said you had 50 60 Beta users out there can you say Something about like what are they Actually doing with the software to help Well do their thing Yeah so the initial Um what what we've had them be doing Over the last uh six weeks is uh We've Provided them with a kind of a Structured curriculum where they're Going through uh tasks that we think are Um that we know that they're they're Kind of very common in the industrial World and so you think about uh moving Objects around so it's kind of pick and Place type tasks as well as we think we See a lot of use for what you can think Of as like machine tending where you're Taking taking an object and presenting It to a machine to have some process Done to it and so we've sort of Canonicalized those sorts of Applications and we our beta users are Really going through the steps of Building applications that look kind of Like those things and we're getting Feedback from them to be able to improve The software so that they can go from There all the way to a deployment of an

Application like that and we feel like There there's a lot of work that can be Done with the robots that are available Today that look kind of like those Applications that people aren't able to Do yet today right not an economically Viable way because it it's just too Difficult to program it and so people Don't put the effort in to program it And instead they'll uh you know they'll Do it by hand they just won't do the Work at all yeah I mean maybe something To say as well is the 50 to 60 that are In the program at the moment or in our Group we've picked out of we had to sell Over a thousand companies apply Uh I'm 50 and 60 are in what we call our Private beta right now actually from October we're going to open it up more Fully so we'll invite more of that a Thousand in but also to be quite honest At that point we'd also love to invite Other people in if they want to actually Try so if you have a burning application You want to try out Um even if you want to build it in Simulation first just to play you know We're here and we also have a set of Labs actually in Munich and in the Bay Area where if you want to deploy your Application and you don't have robotics We have it yeah we'll you can actually Deploy your simulated Application into a real physical cell as

Well and see it running in real time if You want to do that too so if you want To play or you have or you have a real Task you want solving you should come And try as an October yeah I mean in the Abstract those are great answers can you Say something specific about like what What are actually being done sure maybe Let's give an example we um when we we Recently showed um a work cell running In in automatica in Germany it was Basically taking a a phev Um uh module and actually assembling it To a engine frame but it was using Basically you could basically assemble Screws nuts uh in a sophisticated way Onto the frame move that frame and also Disassemble it but then we actually Showed if you move any of those Component parts the whole system could Config itself in real time and why is That it's because we're using things Like vision and touch as well in order To adapt the system with a with a closed Information Loop if you like foreign and I think that was pretty powerful but we Actually showed it running in a process Flow simulation that somebody had Designed then running in reality on the Hardware simultaneously so you can Actually see both things running at the Same time but that was actually quite a Complex assembly set of tasks which Actually to be honest even if you ask

Anybody robotics today even doing Something like a a a a a screen task if Something moves it's actually quite Complex to adapt to yeah yeah Yeah I love that you mentioned that you Have a bunch of labs uh all over the Place and earlier this year I was in the Google robotics labs and I saw a lot of Really cool things in demo unfortunately Most of the things I saw demoed are no Longer around there was a bunch of that Got shot down uh why do you feel you Guys will be different within the Google Umbrella We're already different than that we're Actually already a separate entity so I Don't know if those you know understand Alphabet uh or there's companies like Waymo which is actually a smaller sister Company to Google we're in effect the Latest version of that we're actually Already a separate company and that's Because actually alphabet we validated Enough in the real world already with The real customers that we've actually Become a separate entity so that's for One thing so we're actually protected That way we have our investment Um also it's because we actually have Real customers actually trying our Software and we actually have a real Product as well already out there Um to be honest in terms of those other Uh companies they're also still some of

The some of that talent and learning is Still part of the Google ecosystem and Actually our ecosystem too so there's Great learning I think there yeah no I Get that it feels like a lot of the People who were involved in that were Also involved in layouts and stuff Though over the last couple of years and Like I the reason why I'm passionate About this is I want these two projects To exist and I'm like come on Google Okay come on Amazon leave Amazon Alphabet leave them alone Um do you feel safe in that do you feel Like you're going to continue to receive The support you need to continue We do I mean we we have uh we're we're Very well supported and well resourced And well funded I think um as you know We actually did two Acquisitions Recently including open Robotics and Vicarious recently and I think that's Very much a commitment to the vision and Mission we have that actually alphabet Has supported us in that Um also you know we we very much are Both entrepreneurial in our respect We're very much like a one-on-one bet Where we put our time and energy and so Far Uh I've never done a a business where That hasn't kept growing right and Brian For sure the same yeah and I'll say that It was you know when we you know I was

Running we were running open robotics Independently for 10 plus years and so Why would we think about making a change And and committing uh to join another Company in particular intrinsic and one Of the things that really made the Difference for me was that it was it was Not just a like a robotics research Project within a big tech company like Wendy was saying it's actually it's a Separate entity so it's a group of People who have decided to take this Jump to to leave the mothership and and Go all in on robotics so we're gonna you Know the way I think about it is we're Gonna live or die in robotics there's There's not a there's not a path where We just say well that was fun we're you Know let's go back to doing something That alphabet does uh we're it's there's A level of commitment there both from The people that are involved and also Institutionally from alphabet that I Think is has given me the really strong Confidence that that we're gonna we're Gonna have a shot at it right it's up to Us to actually make it work but I think We're gonna I think we've got a really Honest shot I feel like across the Industry people who are relying on Roz They were like some people like yeah now It's very well resourced and some people Like oh God what's gonna happen next It's funny that you um that you brought

That up like what what is the what is The thing that is possible now that Wasn't before so the you know if if You've used Ross and I hope you have and Robots that you've built Um perhaps one of the things that was Never really obvious was that the you Know the core team that was really Maintaining and developing that software Uh we were you know we were able to do That by uh bringing in uh basically like A you know Consulting customers who Would who would support the development Of it and that if you've you know if You've ever run that kind of business It's it can be actually quite Challenging to ensure like ongoing Stable support and I think it's I'm Really proud of the way that we were Able to make it work Um but what we saw was that there was we Needed to have a step change in order to Really uh take that development to the Next level and be able to really provide What that the whole robotics Community Needs in Ross and that I mean that's Really what motivated the the decision To merge with intrinsic honestly was That it was a an opportunity for us to Be able to have this like to turn up and Really commit over the long term to Supporting all of that open source Software and and to do it not out of a Sense of like altruism or charity right

We're not doing this because We're not just doing it because we think It's good for the world we're doing it Because we're also relying on those Tools ourselves and we're building our Own really scalable business on top of It yeah I think it's it was important for us That you know Brian we really care about The open source Community I mean for Those who've been devs in the past it's Those communities that knew me drive Step change in Industry usually so we we Thought about it very carefully Intrinsic had already been contributing To Ross itself anyway as part of the Community Um and so when we were thinking about Actually merging together you know we Thought both ways about how we're going To protect that community so we actually Continue to invest in it in fact Actually there's more resources now that Are dedicated with intrinsic Specifically to the open source I mean To be quite honest in the future we may Well open source more ourselves into the Ross Community from intrinsic as well we Think it's a symbiosis there I also Noticed a step change and disrupt the Whole industry you cannot do it on your Own and actually the open source model Works very well but actually a whole Ecosystem model works very well and I

Felt that Brian and his team had a super Strong DNA understanding of that but What joins us together is we genuinely Both actually wanted to democratize the Accessibility I've always been about Democratizing technology and all the Businesses I've done and you know Brian's put his whole career on betting On that so we're in it together because We actually really want to make it Happen which is back to your question About why we believe this is going to be The one that actually shifts ecosystem For on behalf of everybody else you know The whole world still runs on GDP I think you know we can't eat the Metaverse I think there's an element of We actually do need to think about Sustainability and production Um with declining Birth rate Etc how are we actually going To deal with that in the world in this Next period yeah no absolutely you Mentioned that you might consider open Sourcing more of the of the software You're developing yourself can you say Anything about like which parts or which Direction you are thinking there Well I think that the you know what you Don't it doesn't it doesn't make sense To just like take a random piece of Software and throw that out into the World you actually take some Consideration to think about is you know

What's the value that this is going to Provide and uh you know can I can I Provide it in a way that people can Really make use of it and can I commit To supporting it right that you know Kind of bad examples of open sourcing Things are just to you know you take Some code toss it over the wall and go On your go on your Merry way right so Um when we think about what makes sense To be open sourced we really want to Look for the the things where we've Produced something internally that we Think has it it's really going to be a Game changer for folks out in the world And it's going to be reusable right it's Not something that's specific to our Software stack but rather it's something That people can use in a wide variety of Settings and you know they're they're Going to be we don't want to say Specifically which pieces uh we would we Would put out there but we're starting To do that analysis right now and think Really carefully about which which are The ones that we think would really be Valuable to the robotics Community Generally and also the Ross community in Particular yeah so this panel is Specifically kind of edging towards Education in in robotics and we've Talked about no code I think I can wrap My head around that in this context of This like what does low code mean and

Like what's like what languages do you Have to learn in order to program those Parts of it The no code level it's obviously sort of A graphical user workflow interface Which you you know some of you will have Seen in other types of software at the Code level we're trying to it's targeted More at a sort of python level developer So almost more like standard scripting Language so we want to make it very easy To use Library based Um an an environment that actually Python develops would be used to so that That's the sort of level we're aiming at But we're giving them access to the Skills and functionality in libraries of A more sophisticated for example an AI Machine learning expert would have or or Even a strong roboticist would have We're also trying to abstract away Things like you know traditionally with You know OT environments you'd be having You'd have to understand more around Lower level controls plcs we're also Trying to abstract that away too so you Don't have to become an expert in that As well Yeah and so the if somebody is sitting In your audience now and they're like Hey I want to get into robotics like how Much do I actually need to learn where Do you begin So I mean if honestly what I would say

Is to I would point you to Ross as a Great starting point there are fantastic Uh tutorials and uh curriculums that Have been put together around the Ross Ecosystem that can introduce you to Simultaneously the the concepts that you Need to understand about robots like you Know how do you interpret sensor data How do you develop a control system what Are the kind of the fundamental uh Concepts that you need to learn but then Also specific tools that allow you to Implement those Concepts and as well as Kind of existing implementations that You can just exercise and so and a lot You know nearly all that stuff can be Tried in simulation so that is going to Be a great place that is going to it's a Great path in that lets you kind of get Used to what it means to program a robot Actually program it run it in simulation And that and that has there's a lot of Breadth in that Community as well it's Worth mentioning that while um you know We're talking a lot here about kind of Industrial collaborative robots robot Arms that are are you know doing work That it you would be kind of familiar to You if you were looking at like a Warehouse or factory but there's much Much more that's going on in the Robotics Community you know you just Heard Mark and Gill up here talking About a much broader variety of Robotics

Applications and we see some of the same Software like Ross used by groups like Them but we see it used by you know NASA Has the Astro b as a free-flying cube That's about this size they've got three Of them on the International Space Station they've got fans on all sides And they fly around in the microgravity Inside the space station that runs Ross That runs the same software that you can Use to program you know a wheeled robot In your factory or or a robot arm and so I think the the key thing to point out Here is that if you start to learn those Concepts they're really transferable There's there's you can go from uh you Know kind of toy examples up to in fact Quite sophisticated systems and and the Same tools and concepts are still there And I feel like for the longest time if You're a Automotion engineer or a Automation engineer in one kind of Um Bubble it was very hard to transfer That skill set to another because the Languages are different all the things Are different and are we now in a place That is possible to abstract it far Enough that the skill becomes truly Transferable I think it I think we're we're getting There and that's you know if I think About what we're doing with at intrinsic We're really uh in particular starting With bridging that gap between uh what's

Been historically the the realm of you Know strictly automation work and the Broader like basically bringing in Modern robotics Concepts that are in Fact transferable I think that you know The if you think about where robots have Been used over the last 50 years it's Mostly been in these kind of really Strict rigid automation settings and That's been it's been a great place to Get robotics started but also they Tended to develop their kind of own Silo Weight of doing things and in parallel The broader robotics industry has Actually Advanced tremendously in terms Of perception and control and planning And so we're really bringing that sort Of modern technology that's available Bringing it to bear on these kind of uh Long-standing older problems Abstraction as well with putting almost A layer of business logic or task flow On top of that you can see the step Change from there will also come with Things like genitive AI so gen AI right Now once you know that if you can Abstract out that business flown and uh Like let's say task building logic At some point it'll just become a Prompt-based interface you won't even Need a process flow or a graphical user Interface or even ultimately python Coding to be honest and I think you know It's one of the things our R D team

Unusually even though we're just a new Product is already working on because of The step changes are happening in the Rest of uh you know the tech sector Right now we're adapting to that so I Think this is actually very timely that The abstraction we're actually doing is Actually going to enable a very quick Shift to what's going to be happening in The next around artificial intelligence Yeah well I've seen some demos along the Lines of where you can talk to a robot And say hey there's a kind of coke Please put it in the bin and the robot Goes scurrying around so it has like a Pathfinding picks it up knows where to Put it all it knows where to look and I'm curious like how far are we away From being able to do that in a more Industrial setting like hey fasten this Thing to that thing What I was referring to it starts to get To that because one of the interfaces Could be a natural language processing Interface taking that business logic so For example if somebody had did what I Described earlier of the uh the sort of Engine assembly task Um there was already a workflow with Photographically that runs that whole Process so you can imagine it wouldn't Take much to actually use voice command Actually to create that flow or to edit That flow but as Brian said because of

The the computer vision and the uh false Feedback sensors also them that can have A real-time response an adaptation Um you could literally say look um can You can you move that piece over there And actually can you then re-run the System Um so you know I think for the Roboticists I think you would say this Wrong as well that that seemed to be Science fiction for a very long time Yeah but actually it's now in the Realms Of the next firm well I'm gonna I'll Make a statement next five years rather Than 20. yeah Yeah beautiful I think that's a great Place to end thank you all for coming Thank you both for joining us [Applause] [Music]