Christoph Loytved of COMPREDICT

0:00 Introduction to AutoTech Detroit

Today in The Garage, we’re recording live on the show floor of AutoTech Detroit. Our guest today is Christoph Lloydfed, product manager from COMPREDICT.

COMPREDICT is helping lead the way with virtual sensors. And in the episode, you’ll learn what is a virtual sensor and how it provides upstream and downstream benefits for OEMs and consumers alike. It’s almost the perfect application of software defined vehicles. Let’s hear about virtual sensors from COMPREDICT.

Let’s go.

0:38 Guest Introduction: Christoph Loytved

Welcome to The Garage. I’m John Heinlein, Chief Marketing Officer with Sonatus. We’re filming live at AutoTech Detroit so you can hear some of the background noise. My guest today is Christoph Loytved, product manager from COMPREDICT.

Christophe, welcome to The Garage. Hi, John. Thank you. We always like to start by getting to know our guests.

Tell us a little about you and your background.

0:55 Christoph’s Background and Experience

Sure. So, I studied at TU Darmstadt. I did my master’s over there, a little bit of time at UBC as well studying. Then I went, for a job at COMPREDICT, R&D engineer. So I developed virtual sensors in the first years. I think it was around five years. And now I recently switched to product management, now being responsible for tire wear and brake wear virtual sensors.

That’s great. And you’re based where in Germany?

It’s, in Frankfurt. Darmstadt, actually.

That’s great. So, we we always ask our guests to tell us a fun fact about them. Tell us your what’s a fun fact about you?

Alright. So for me, I love dancing, actually. So, specifically salsa and, a little bit of West Coast Swing recently.

So really, this is something I like to do wherever I go. So actually, here in in Detroit, I also went dancing on the weekend. That was really nice. It’s a little bit like an in a different language. I don’t know if you know that.

It is. I mean, it’s and it’s also universal. Right? You don’t have to speak the same language, but you can still dance.

This is right. This actually, this is the fun. I can go wherever I like. I went on Hawaii as well.

I didn’t know they were dancing there. Yeah. It’s awesome. I tell you.

So I always try to come up with a fun fact back to our guests. I think my fun fact would have to be my wife and I, years ago, we got married. And many many couples, you know, learn a kind of a dance routine. So we actually took classes. We learned a dance routine, and we did a, did a dance in our in our wedding. So there’s my there’s my fun fact.

Awesome. That is awesome.

2:25 Overview of COMPREDICT

So tell us about COMPREDICT and your role.

Sure. So COMPREDICT was founded around, 2016. Sorry about that.

It was a spin off of the TU Darmstadt, so mechatronics.

Back then, we are mostly concentrated on damage estimation gearboxes.

Nowadays, we really focus on virtual sensors. So basically making the most out of existing data in the car, really, ensuring that we have, better quality data in the car, replacing hardware sensors or giving additional, sensors information to it. Yeah.

And making virtual sensors a gold standard and

3:07 What is a Virtual Sensor?

So you you have to start for our listeners to help us understand what the heck is a virtual sensor?

This is right. This is a strange concept for some people. Yes. Yes. Explain it.

We get it all the time. We get it all the time, John. So, really, how it works is it’s a piece of software running in the car.

It takes all the sensors which are existing in there. So there’s a lot of like wheel speeds, accelerations, and on.

So so real signals, hardware based signals.

Exactly. In the car. Yeah. And it smartly combines them to to replace hardware or to add additional functionality to it. So it’s it’s basically a mix of physical models and machine learning, which makes, like, your your your data that you already have more valuable than the cloud.

So deriving different data points from data you already have into new new ways.

Yeah. You could say that.

Really, really interesting.

Really interesting.

3:58 Applications of Virtual Sensors

And so what what are some of the applications of can you give us some examples to make it easy to understand?

Sure. Sure. Let me give you two examples of that. Alright? The first one, headlight leveling, for example.

And headlight leveling, you have, like, two to four sensors in there measuring the ride height. And with these kind of, like, sensors, you can adjust calculate the pitch and then adjust that light. With our virtual sensor technology, we can get rid of those hardware sensors measuring this ride height and just do fully virtual or partially virtual as as our customers like and thereby save costs for the company. Okay.

What’s another example? The other example would be tire wear, for example. This one is not is not replacing hardware, but it’s actually giving additional information into the car. So for tire wear to to my knowledge, there is not really any hardware measuring, at least not in serious cars, serious production cars.

Yes. And, basically, what we’re doing, we’re giving this information to the car and thereby giving our customers the opportunity to know when tires need to be replaced.

5:00 Benefits of Virtual Sensors for OEMs

That’s fantastic. And so so thinking about the benefits to OEMs, you touched on it a little bit, you know, cost savings. But what are some of the the motivations to shift from hardware-based sensors to virtual sensors?

Yes. We had talked about the the the hardware replacement, which I think is the obvious case, saving costs. But then, like, really, like, getting also additional information into the car, for example, like getting back to the tire wear algorithm.

It really helps, to understand when a tire will be worn, to target customers really specifically to go back to the OEM-branded workshops. Right? So you can really have the spare part business increased. We’re estimating around 180 Euros over over lifetime for a car. So that’s actually good money.

That’s interesting. So so upfront cost savings from sensor you don’t have to put in the vehicle, downstream savings from, I would say, value added things you can do Mhmm.

That maybe benefit it probably benefits the consumer as well from a safety perspective. Of course. Benefits the, the OEMs, potentially ongoing revenue models.

6:01 Data Utilization in Software Defined Vehicles

It’s interesting. As I reflect on software defined vehicles and we talk a lot about this from many different aspects of the podcast, there’s so many, additional signals being captured, additional data points being captured from vehicles today.

But companies, OEMs don’t always know how to take advantage of that. And so either in some cases, that data is just not used or perhaps it’s captured but put in a giant pile that is never…so I think this it seems like what you’re doing unlocks the opportunity to mine that data in new ways. Yeah. To create new value. And I imagine the great part about virtual sensors is you can add them after production.

Yeah. Everything’s possible with these. Right? There’s just a multitude of possibilities that we’re having here. So in in in production, like, you you could save the the hardware, but afterwards, even if the hardware breaks, right, post production, you could say, like, do we replace it with the hardware, or do we actually put them in the virtual sensor that COMPREDICT is doing? So there’s multiple options there as well. Doesn’t need to be only on production.

This really depends.

7:07 Market Adoption and Investments

So, tell me about, your production adoption. Yeah. How how wide is is your market adoption so far?

Of course. So one thing is talking about dreams and stuff like this. It’s always comes down if it’s really used. So we’re quite happy that we can actually talk about some customers that we have.

That’s not always the case. Before I would talk about this, perhaps let’s talk about one big factor which enables us, which is Toyota, which invested recently into us. So it’s it’s not only us believing in this kind of mission, but it’s actually also Toyota Motor Growth Fund, which invested into us. But then, of course, customers are more important.

And one one person or one group we can talk about is Renault Group. So really just, like, all the brands under the Renault Group like, Alpine, Dacia, they are all adapting our virtual sensor technology for tire and brake wear. So we’re estimating by 2030 around ten million vehicles.

That’s fantastic. So across the Renault Group, adoption of your technology as the first application for tire wear.

That’s huge. That’s huge. Yes. Congratulations. Thank you so much. And let’s see let’s see if we can, like, put our numbers soon on our website as well as you do, you know?

Well, thank you. Thank you for the kind words. Yeah. Synodos is in, depending when you watch this, four million vehicles today.

But by the time you watch this, probably five or many more millions of vehicles. So we’re we’re growing rapidly. But congratulations to you. That’s a great achievement.

Yeah. Thank you.

8:34 The Future of Virtual Sensors in SDVs

So if we think bigger picture, it it occurs to me that these virtual sensors are really a great application for software defined vehicles.

What are some ways that you think that that SDVs really are a perfect match for virtual sensors?

Well, there’s actually a lot of dimensions we could talk about. Right? One way for us is really, like, we’re making efficient use of this all this data. Like, we’re running our virtual sensors. We so we have the ability to run virtual sensors easily in these cars, crunch the data there, upload the data, which is only relevant and not all this this the big data amounts.

Actually, the integration we own…

That’s a really interesting point.

So so think about it in terms of instead of having to send all the data to the cloud for processing, the virtual sensor in the vehicle can derive the data and send the derived data, which is probably far smaller…Of course…

Than the, than the the source data.

This this is how it should be. Right? Like, we should only send relevant data and not all all the data to the cloud. And, as of now, for some information, you have not simply not the possibility to send it from the car. But with our virtual sensors, we can create it in the car and then send to the cloud, like the tire wear, for example. There is no information on tire wear.

So that’s a huge potential cost savings for data management and and data manipulation.

Yes. For sure.

That’s great. You were mentioning integration. Sorry. Yes.

Thank you, integration is obviously also a very important point for us as we’re trying to build really scalable software, which runs in many cars.

Right? So for us, we don’t wanna go to through all these, like, points of, like, integrating in this car and with this kind of software interface and this kind of software interface. We really want to have a standardized way to access this kind of data, to put our software. Optimally for us, we just plug the data to the virtual sensor, and it works.

So we believe that in virtual with the SDV kind of developments, there will be a clear way of implementing our third party software into OEMs cars without also the hassle of OEMs not knowing how to do that the best because it’s standardized. Right? So this is really big for us.

Yeah. And that really speaks to the broader question of SDV. You know, there’s been a feeling over the past few years that, SDV was a buzzword, but people were wondering, was it real, or is it just an interesting idea?

What I hear you saying is that this, this and and things like it are real applications that can make SDV really valuable in in creating new applications.

Yeah. Yeah. I agree. I think we are one of those use those of one of those many use cases which can make it really helpful, which can make make most of the SDV that we’re we’re discussing. And it’s really an application case, not just a thought about what we could do theoretically, but actually something we can do. And we know already it works.

That’s fantastic. So, Sonatus, one of our claims to fame is, high resolution data capture among many other things.

So I it feels like a great synergy with you. We we’re really excited and look forward to, continuing to work with you in the future.

We’re thrilled for your, your great success with Renault and look forward to more customer adoption for you coming up.

Thank you so much. Yeah.

11:49 Conclusion and Closing Remarks

Yeah. Hey. Thank you for joining us on the podcast.

Thank you. Thank you, John.

If you like what you’re seeing on this episode, please like and subscribe to see more like it, both from episodes here at AutoTech Detroit as well as for our home studio. Thank you again, and we’ll see you in another episode very soon.