The Sputnik moment of automation?

00:00:00: [MUSIC]

00:00:05: >> Hello everybody and welcome to a new episode of our Tech Podcast by Bosch Rexroth.

00:00:09: My name is Robert Weber and my guest today is Thomas Fechner.

00:00:12: Thomas, welcome back to the podcast.

00:00:14: >> Hello, Robert. It's always a pleasure being with you.

00:00:16: >> Thanks a lot.

00:00:17: We want to talk a little bit about platform-based engineering.

00:00:22: You showed case the deep-seek example a few weeks ago,

00:00:27: and you said that this is a Sputnik moment for us.

00:00:31: What is the Sputnik moment now in automation?

00:00:34: >> Of course, it comes from the software domain.

00:00:36: I like the example because deep-seek showcase how strong you can be

00:00:42: utilizing platform-based engineering compared to the traditional way of working.

00:00:47: >> What does it mean platform-based engineering now for the automation business?

00:00:52: >> The automation basically is still working very traditional,

00:00:56: and if we look how the productivity in software engineering is increasing,

00:01:01: then we are very jealous and like to transfer

00:01:05: these success factors towards the automation domain.

00:01:08: Platform-based engineering is the method what is really progressing a lot in the software engineering.

00:01:14: There's a gardener figure where you can see that we are in the middle of the trend of going

00:01:20: from 40 percent platform-based offerings towards 80 percent platform-based offerings in two years of now.

00:01:26: >> This is the method which is making the difference in software engineering.

00:01:31: >> But what is it? Can you describe the method?

00:01:33: >> Okay.

00:01:34: >> What does it mean to do that?

00:01:36: >> Yeah. The base of platform-based engineering is that you have

00:01:39: the standardized platform on which you develop your features.

00:01:43: So, this platform must solve the-

00:01:45: >> So, it's your control-X automation.

00:01:48: >> In the automation domain, it's the control-X OS,

00:01:50: what we are delivering. It's an operating system.

00:01:53: What is basically managing the data exchange,

00:01:56: what is managing the decoupling of hardware.

00:01:58: So, hardware software gets complete decoupled.

00:02:01: And what is managing, of course, all the distribution questions.

00:02:04: So, how can I commission the system?

00:02:07: How can I update the system?

00:02:09: How can I access the system?

00:02:10: So, these things are solved in the middleware.

00:02:12: How is communication working?

00:02:14: So, based on this-

00:02:15: >> Is this a new layer?

00:02:17: >> This is a middleware layer.

00:02:19: And it is similar to what we experienced on the IT side

00:02:22: when we make the switch from the mainframe

00:02:24: towards the personal computer.

00:02:26: Yeah. All of a sudden, the operating systems appeared,

00:02:29: what we all know very well with all these revisions.

00:02:32: And the same happened, of course, for the mobile phones.

00:02:35: The first mobile phones, maybe you remember,

00:02:37: were really embedded systems.

00:02:39: So, software hardware extreme deep,

00:02:41: and the companies were even making-

00:02:44: >> Very successful.

00:02:45: >> At the beginning.

00:02:46: >> At the beginning, yes.

00:02:47: But then we had a really explosion of innovation and shift

00:02:51: while we got the iOS and the Android operating systems,

00:02:55: where a community could contribute all of a sudden

00:02:58: with a lot of new functionality and ideas.

00:03:00: >> But this is interesting because you mentioned iOS and Android.

00:03:04: And these are two different worlds.

00:03:06: We have an open source Android system,

00:03:08: and we have a closed system.

00:03:10: So, what is your favorite?

00:03:11: Is it a proprietary system like iOS,

00:03:13: or is it an Android more open source?

00:03:16: >> The interesting thing is,

00:03:18: also on iOS, you are doing a platform-based engineering.

00:03:21: It's just the way how you commercialize it.

00:03:24: It's a bit different.

00:03:25: But with the iOS, you also have built a platform

00:03:28: where you have a community which is contributing.

00:03:30: The community is managed different in both ways.

00:03:33: The monetization is working different.

00:03:35: >> But let's talk about the automation community.

00:03:37: They are not focused when it comes to open source, right?

00:03:40: So, open source is a totally new playing field for them.

00:03:45: They need to learn what does it mean.

00:03:47: There's no free lunch when it comes to open source, etc.

00:03:50: So, what do we need to learn in the automation business

00:03:53: when it comes to platform-based engineering?

00:03:56: >> Yeah, of course, utilizing open source,

00:03:59: this is what everybody is keen on

00:04:00: because you have not to redevelop it.

00:04:03: So, you utilize value which is already created

00:04:06: again and again and again.

00:04:07: By this, you are scaling and you get productive.

00:04:10: Of course, you have then the responsibility

00:04:12: towards the customer to maintain it.

00:04:14: >> Yeah, that's your business model then.

00:04:17: >> Exactly.

00:04:18: So, utilizing open source code means on the other side

00:04:22: the responsibility to maintain and to keep it updated.

00:04:25: And this is what is very often forgotten.

00:04:28: It's always thought as it's for free.

00:04:30: And of course, by this, I also not willing to pay for it.

00:04:34: That's for sure not the right way.

00:04:35: >> Yeah, but there's a business model

00:04:38: behind open source projects, right?

00:04:40: A lot of open source projects has also a business model.

00:04:43: There's a company like Red Hat

00:04:44: and they are using open source project

00:04:47: to make money out of it.

00:04:48: So, there's a business sweet spot.

00:04:50: >> Yeah, of course, of course,

00:04:51: but you always add then things to the open source functionality.

00:04:54: >> Exactly.

00:04:55: Is this the way for the automation business,

00:04:57: totally focusing on software-first developments?

00:05:01: >> So, software-first thinking is for sure

00:05:04: one of the paradigm shifts the machine builders have to learn.

00:05:07: Very often, it's the question,

00:05:10: what does this mean, software-first?

00:05:11: Does this mean I need an engineering chain up front?

00:05:14: Do I need to virtualize everything?

00:05:16: No, that's not what I see and understand

00:05:19: when I'm talking about digital-first,

00:05:21: but it's the obligation that you have to think

00:05:23: about your automation architecture.

00:05:26: Because at the end, you like to get fast and productive

00:05:31: in building your machine.

00:05:32: You need to think about the efforts for integration.

00:05:34: And this architecture, you have to define up front

00:05:38: before you start.

00:05:39: And this has to do with the selection

00:05:41: of the right automation.

00:05:41: >> So, it's a software-defined machine.

00:05:43: >> A software-defined machine at the end.

00:05:45: And this leads you, in my absolute confession,

00:05:49: it's the right word, I think, to an operating system

00:05:53: which you can deploy on different building blocks

00:05:55: of your machine.

00:05:56: It can be different hardware vendors, even,

00:05:58: where you work with the same operating system,

00:06:01: which enables you later to integrate everything very effective.

00:06:04: >> Your colleagues from the Bosch Automotive,

00:06:07: they established the slogan "software-defined week", right?

00:06:10: So, you are going the next step, software-defined machines.

00:06:14: >> Yeah, for sure.

00:06:15: I think we are already very far in this,

00:06:17: that everybody is clear that software makes a difference

00:06:20: in machine building already.

00:06:22: The question is, if we are doing it very professional,

00:06:25: and in the end, the toolbox we are utilizing is a very aged one.

00:06:30: >> From a machine-builder perspective,

00:06:32: what does it mean now for him?

00:06:34: Is he now capable to sell also digital products?

00:06:40: As an add-on to his machine, to his hardware?

00:06:43: >> Yeah, if a machine-builder is making the step towards

00:06:47: platform-based engineering, then he gets the capability

00:06:51: to add functionality later during runtime of the machine,

00:06:57: which is coming by new software functionality.

00:06:59: A very good example is what we get now

00:07:02: from the artificial intelligence side.

00:07:04: A lot of analytic features are coming on top,

00:07:07: and how to deploy them on an existing machine,

00:07:10: which is very close in a proprietary solution.

00:07:13: >> But my feeling is that they are not able,

00:07:17: or it is very difficult for the sales guys,

00:07:21: to sell digital products to the customers.

00:07:24: And they are doing something,

00:07:26: maybe we should stop this whole software stuff,

00:07:29: we are still focusing on our machine.

00:07:32: >> Yeah, but you need to see also the customers,

00:07:34: or the end customers, who are operating the machines

00:07:37: they are learning.

00:07:37: With each and every crisis we have,

00:07:41: they see that adoption, flexibility,

00:07:44: fast reaction on changing environments is the key.

00:07:47: And this always leads you towards the requirement

00:07:50: of the machine to get more flexible and more adaptive.

00:07:53: >> If you compare it to the, for example, automotive industry,

00:07:57: the customer also wants software edits,

00:08:00: they want new software ads,

00:08:02: and that's a problem for the maybe European,

00:08:05: German car manufacturers,

00:08:06: because they can't offer so many updates

00:08:09: when it comes to software and to new functionalities in the car.

00:08:13: Maybe it's a good sign that we now get a lot of pressure

00:08:17: for the machine building companies to offer things like that.

00:08:20: >> Absolutely, and we see that there's really a shift

00:08:24: in understanding, besides software first,

00:08:27: what you already mentioned,

00:08:28: there's a second paradigm shift.

00:08:30: which is very important. This is true openness.

00:08:32: What does it mean? True openness? So, competition or what?

00:08:35: Yeah, at the end, it's even competition or realizing win-win-win situations.

00:08:39: And that's too nice. Win, win, win, win. Everybody wins.

00:08:43: Everybody wins. Everybody wins means at the end, you are not winning all.

00:08:48: So you're ready to share.

00:08:51: You have to share. So that's that's maybe the better explanation.

00:08:54: And this is something what we are not so used in the classical industrial space.

00:08:59: So this is always being proper. I want to win everything.

00:09:02: Have everything under control. It's my IP.

00:09:05: We even secure ourselves a lot with IPs.

00:09:08: And this is what you have to overcome and work really towards ecosystems.

00:09:12: IPs for the weak companies?

00:09:14: No, no, it's not for the weak. It's the Elon Musk philosophy.

00:09:17: Yeah, yeah.

00:09:18: I think we have to differentiate on the hardware side.

00:09:21: IP still has a really strong meaning also for us on the software side.

00:09:26: A different understanding. That's all about speed.

00:09:28: Yeah. And true openness.

00:09:30: So we need to learn to work together with the GitHub guys.

00:09:33: Nobody knows these guys. We're working there in this project.

00:09:37: It's very difficult for traditional machine building companies or for a traditional company

00:09:42: like Bosch to trust these guys in this project.

00:09:45: Yeah, that's true.

00:09:47: So in the end, we have really the opportunity for huge communities to contribute.

00:09:52: But this you can somehow control with whom you are working, of course.

00:09:57: And let's first start in your own company to get all your engineers on board

00:10:02: to contribute on a platform to make code, which is reusable.

00:10:06: Second step is go into a community where the people know the domain

00:10:11: and you already can broaden a lot.

00:10:13: Invite startups or young companies who have a strong innovative offering.

00:10:20: What usually has the problem to get integrated in established

00:10:24: and proven automation environments.

00:10:26: So all these things you can start and if you if you gain trust in this

00:10:31: in this open ecosystem work and in the work in communities, then of course,

00:10:35: you can open really towards the whole world to GitHub world.

00:10:38: Yeah, I think a good example is Bosch Rexford is very strong when it comes to

00:10:42: Ross. You invested a lot of time and code and people went to the conferences.

00:10:49: I always meet Bosch Rexford guys there.

00:10:51: So Ross is a very big topic for you, right?

00:10:54: And that's I think it's a good example how to to be a part of this community,

00:10:59: not like I'm the big Bosch.

00:11:01: I want to show you what we can do.

00:11:02: So maybe to collaborate with smaller companies to give hardware to share hardware.

00:11:09: And I think it's very interesting what you're doing in this Ross communities.

00:11:12: I think we can transfer the learning from there.

00:11:15: Why we are engaging in Ross is because robotics and Ross is the operating system

00:11:20: for robotics.

00:11:21: Robotics is an integration art.

00:11:24: I would even say and the differentiating functionality comes always from

00:11:29: universities or from young companies and the traditional ones.

00:11:32: They are looking for industrialization and make it very reliable in also all

00:11:36: the supply chain questions and so on.

00:11:38: So the need was very fast there to have an integration platform and where do

00:11:44: they head to towards an operating system?

00:11:46: Somehow we did not make the transfer to a classical machine building.

00:11:50: And this is what we offer with our automation platform.

00:11:53: So the industry knows already that a success factor is the standardized

00:11:57: middleware which is integrating.

00:11:59: You mentioned GitHub.

00:12:01: When I go to GitHub, I don't see any IEC 61131 code there.

00:12:08: So do we need to learn to speak new programming languages in the industrial

00:12:15: sector in the automation business?

00:12:17: Yeah, of course.

00:12:18: And especially if you look what the students have as knowledge when they

00:12:22: come from the universities, usually they didn't learn the ladder logic and how

00:12:27: we did it traditionally.

00:12:28: So this is also one of the tasks of the operating system you offer that you

00:12:34: can enable the users later to use any language encoding their...

00:12:39: But you also support the old stuff.

00:12:42: We also support of course the old one.

00:12:43: But all the apps you let run on the operating system can be coded in

00:12:48: different languages.

00:12:49: It can be a Python, a C++ even in combination.

00:12:51: It's no issue.

00:12:52: Okay.

00:12:53: I want to come back to the middleware one step back.

00:12:56: The middleware.

00:12:57: Is this middleware also capable to provide updates, to provide dev ops, to

00:13:04: provide updates when it comes to cybersecurity, etc., etc.

00:13:08: So it's also a communication layer or can you go a little bit more in details?

00:13:12: Yeah, of course.

00:13:13: In our point of view, the middle layer has the tasks first to decouple hardware

00:13:20: and software.

00:13:21: The second topic is to organize the data management between devices as well as

00:13:26: between the different functional software blocks which come on top.

00:13:30: Also to manage the software blocks in the runtime.

00:13:34: When do I execute which one and how do they communicate between each other?

00:13:38: And then, and this is the important thing, manage all the updates, all the

00:13:43: administration of the content.

00:13:46: And this is what you need in a managed way in your machine, in all the

00:13:50: intelligent devices at the end, if you like to get cyber resilience capable.

00:13:55: And if we think a little bit further, when we talk about virtual PLCs, you

00:14:00: also need these communication layers then.

00:14:02: Yeah, of course.

00:14:03: If you have the operating system, which is solving the topics I just mentioned,

00:14:08: of course, you can also deploy it on an IPC and even higher in the cloud.

00:14:13: Yeah.

00:14:13: So you can make a vertical distribution of your code.

00:14:17: What is nowadays just possible if you make at the beginning a distribution,

00:14:22: what do I solve where, what to deploy where in the open system, we are

00:14:28: thinking you can deploy at the end.

00:14:30: You can first develop your automation code and then you can think about where

00:14:34: does it need to run to have the right real time capabilities where you

00:14:39: need a fast response of your hardware.

00:14:42: Of course, you go on an embedded device in the machine.

00:14:45: If it's something which is just needed to be close, you have a 200 millisecond

00:14:50: latency requirements, you go on an HPC.

00:14:53: If it is something analytics, you just update a few times in an hour, you

00:14:58: can go to the cloud.

00:14:59: Yeah, exactly.

00:15:00: At the end, I always ask my guests, the times are not easy for the automation

00:15:05: business.

00:15:05: What are your hope for the whole community, for the whole automation community?

00:15:11: Yeah, first of all, I'm absolutely convinced automation has a great

00:15:15: future because automation is a mega trend.

00:15:18: We see even that it's accelerating, looking on the functionality which is

00:15:22: coming from the IT side and we need to bring it in our system.

00:15:26: And you see new competitors, right?

00:15:28: Of course, also new competitors and usually this keeps us running and

00:15:32: innovating faster.

00:15:33: So we are really, we like the challenge.

00:15:36: Yeah, we are not running away from it.

00:15:37: So that's the mega trend topic.

00:15:40: Another thing is, and this goes a bit deeper.

00:15:42: If you look what happened with the digitalization of our work desks the

00:15:47: last 10 years, you can be a bit desperate because there's no more, a lot

00:15:51: devices there.

00:15:52: So not much business you can do as a device supplier.

00:15:56: It's ending.

00:15:57: So printer business is gone.

00:15:58: Exactly.

00:15:59: This is gone.

00:16:00: This will not happen in machine building because at the end we are always

00:16:04: treating physical goods.

00:16:05: We are adding value to physical goods and this will stay.

00:16:09: So it will be at the end a very intelligent combination of hardware and

00:16:13: software, what makes the production of the future.

00:16:16: And this is what we like to contribute to.

00:16:18: And you have one more?

00:16:19: Yeah.

00:16:19: And the third one is that we really see that a movement starts to move away from

00:16:24: the traditional way to automate.

00:16:25: For example, we offer hackathons for customers which are a bit struggling.

00:16:29: They are, let's try it out.

00:16:31: You say, okay, let's come.

00:16:32: Let's invite the crazy guys.

00:16:33: Exactly.

00:16:34: And bring them together with the software architects and always they go out with

00:16:39: a smile in their face.

00:16:40: And one thing they always say this, we have never expected that we can accelerate

00:16:45: so much with this approach.

00:16:47: Thomas, thank you very much for the pleasure.

00:16:49: Thank you.

00:16:50: [MUSIC PLAYING]

00:16:53: [MUSIC PLAYING]

00:16:57: (gentle music)