The benefits of digitalization for AI automation – Podcast Transcript
Building adaptable, flexible automation system is both a goal and a challenge in the manufacturing world which must train powerful AI solutions to meet these needs. As the industry continues to embrace digitalization, not only will it training and testing of advanced AI systems be easier, it will be a key step in realizing the benefits of software defined automation and the Digital Twin.
To learn more check out the full podcast here or keep reading for a transcript of that conversation.
Spencer Acain: Hello, and welcome to the AI Spectrum Podcast. I’m your host, Spencer Acain. In this series, we explore a wide range of AI topics from all across Siemens and how they are applied to different technologies. Today, I’m joined by Christopher Schutte, Senior Product Portfolio Manager for Robotics AI at Siemens Digital Industries.
Spencer Acain: To kind of pivot from where we left off last time, how do you see the broader impact of everything we’ve talked about in the previous episode? So AI automation and physical AI and all of that kind of stuff benefiting from the broader shift to a digital industry or digital enterprise, such as software-defined automation and PLCs and those type of digitalization trends that have been ongoing in the industry over the last few years.
Christopher Schuette: Oh, yeah, definitely. I agree. We benefit. So I think I mentioned that the thinking and system value is a core principle at our team. And that means we have this fact that the world outside the industrial world is mostly PLC based. But we integrate these novel technologies into new solution types like the AI enabled robotics. And this requires to integrate complex AI models, maintain them, and run them on hardware that will be in place for five to seven years. it’s not a gaming pc that you can put there and okay I’m fine that would be probably not the case it should be something with spare parts support over all these lifetime you need to enable this data flywheel concept and there also the shift in the digital enterprise helps heavily why because there are concepts like industrial edge that helps in secure data and chain exchange Concepts like automation expansion at Siemens makes engineering in machines much more IT-like. That means it coming more from a one-time development of an AI robot cell for one customer that is like tailored for this customer, more to the methods of IT engineering where you think in modularity, where you think in to put things together as a product and not as a specialized solution and next to it also the plc changes its format so it’s a hardware plc mostly today the gray boxes and siemens other color colors at other suppliers and they’re merging now to virtual plc that you can directly install next to your ai models and make it easier to maintain engineer giving all these things in the hand of your engineering and service staff and this all this type of streams that happen in engineering in our edge in also to virtual plc next to new technologies integrating all these things together are a huge benefit and we’re thinking every time i said in the system value to give in blueprint and architecture how an ai enabled robotics solution could look like that works for five to seven years easy to maintain easy to upgrade and it’s secure. Yeah, really benefits. And the last point is the digital trend. So before everyone wants to build this thing up and invest $200,000 or Euro in such a demo, they want to see if this helps. And for this reason, they build up simulations. It’s not a digital twin. I’m aware of that. Simulations, they prove the value as good as possible in the digital world. They build up, reuse code, and then going the real world, they have the digital twin, enrich it. And it’s also the goal that we probably have the SIM to real gap at some time close that we can probably also get the data fly running with cohesive digital twin. Would be great.
Spencer Acain: Yeah, no, definitely. So how do you see this type of AI and automation advancing in the future? You’ve obviously made a lot of amazing strides recently and what you’re expecting in the short term, but… How do you see this improving more in the future? What’s the next step for you guys in terms of improving these robotics and automation systems?
Christopher Schuette: It’s probably an own podcast episode, but giving you some points. We have a point with pick AI where we can enable adaptive flexible robotics and we use their foundational models for segmentation next to the more deep learning traditional neural networks to do other jobs in the product. The next step that will come up is to get advantage of so-called vision language models, VLMs. That is already real. So it’s getting more and more in production. We’re also trying that thing out. What this gives is it gives an understanding of the real world scene to the machine. That means use cases like identified damages in the items stored, search objects, sort them, count them, all the things getting immediately possible. And the L in the VLM just indicates that you can’t prompt all of this by directly prompted with your keyboard or you just make it in the background. And that is amazing. That is one step. This is already happening. The long-term five-year plus goal is to have really a robot foundation model that in range of covering really close to autonomy and the actual architecture where the people bet on our original language action models. This type of AI takes full control over sensors, actuators, and in a high-frequency scanning adapting cycle. They hopefully will generalize all the tasks about the use case around customer sites and settings so different cameras different end of arm tools robots that would be the great hope and with this reducing tremendously any engineering effort one is one of the big consume in money in any solution and bring us all to more and real intent based automation where users just say the or ask the machine what to achieve instead of program how to achieve something. And we are on the way to achieve this vision more and more. It’s an early stage, and this technology is also one of the foundations of designs, machine designs like humanoid robots. Today, the humanoid robots are impressively and a really awesome, mechatronic, beautiful results. but the software in it needs to catch up. It’s not that flexible and it’s not that in a human at Robert today is autonomously acting. We expect that because it looks like human, it should like act like a human, but it’s not doing this at the moment.
Christopher Schuette: Today is mostly also remote controlled and doing very simple task. And we are happy to see them doing the simple tasks at automotive manufacturers, for example. And I’m very curious what will happen if we solve this problem of huge robot foundation models that really brings true autonomy and that is hard to achieve in a reliable manner, industrial grade. That is a very hard problem to solve. And we are there. And then you can adapt these type of technologies. Just talking about inter-logistics yet as a starting point to land and expand this type of to solve problems. But manufacturing have the problem of a high and growing variety of items. They need a lot of adaptive and autonomous systems. As said, we have 40 to 50% of today’s tasks are not automated with traditional automation methods. You can go outside of the building and say in farming, what you can achieve here? What can you do in pharmaceutical? in household at the end it’s especially the situation where machines seeing and need to interact with our real world so it’s just opened up a tremendous new paradigm shift and super curious to be at the forefront of that
Spencer Acain: Yeah, no, I mean, it sounds like we’re just on the cusp of what AI can achieve in the factory on with robotics. So I guess to round things out here, do you have any final comments on AI? Either things we haven’t had a chance to cover today or, you know, what is the broader impact of ai and the manufacturing industry going forward?
Christopher Schuette: Yeah probably so we’re in contact on a daily basis with prospects with customers with partners and what turns out is one thing thatit’s so important that this trust in industry not only in the industry it’s also you know from all personal relationships But my personal belief is behind all the marketing buzz that is at the moment around the AI hype, there will be the time between the AI hype peak and where we reach the point of reliable productivity, where we all need to manage expectations to customers, to partners, stakeholders, and we are all humans. And what is my personal golden rule is providing transparency, what is possible, what is probably risky. It’s all about de-risking things in industry. Wherever possible, I want to give trust because this converts to a very healthy customer relationship and it’s the glue between partners and stakeholders in the industry. That would be just my two cents. It’s great to talk about the technology. If you solve a customer problem, it’s much better. And if you do this in a way to solve the customer problem in a reliable way where people feel, oh, they really take my risk here and help me to de-risk everything and want to make me successful instead of selling me a hype that really brings a lot of credits to you and your company.
Spencer Acain: Yeah, no, I couldn’t agree more. Trust is absolutely one of the biggest issues with AI, I think with generative AI and everything else, making sure you can trust your AI results is perhaps the most important thing of all when it comes to this. That is about all time we have for this episode. So, Christopher, thank you for joining me with those great insights. Thank you very much, Spencer. And once again, I have been your host, Spencer Acain, on the AI Spectrum podcast. Tune in again next time as we continue exploring the exciting world of AI.
Siemens Digital Industries Software helps organizations of all sizes digitally transform using software, hardware and services from the Siemens Xcelerator business platform. Siemens’ software and the comprehensive digital twin enable companies to optimize their design, engineering and manufacturing processes to turn today’s ideas into the sustainable products of the future. From chips to entire systems, from product to process, across all industries. Siemens Digital Industries Software – Accelerating transformation.
