Sensors with Superpowers: Farnell’s Cliff Ortmeyer on AI and IoT

Ruth: Yeah, and we will, talk about use cases in depth a little bit later because that's what I'm also fascinated about is, yeah, how, to see how this abstract topic and the things that were usually only for techies or nerds are now becoming real life solutions to real life problems.

That's what I find also really fascinating. Yeah,

Cliff: I agree.

Ruth: And does AI play a big role in this development? And the rise of sensors.

Cliff: Yeah, it absolutely does. For me, if you, look at where sensors have been used, sensors have always been the core foundation to IOT, right? Any type of sensor. and if you look at where that's where AI is now taking us, having that data.

The raw data from a sensor is great, but now when you have multiple sensors and you have huge streams of data, I have a pulse from a glass break or anything like that, you have to be able to detect what that noise is. Cameras, for instance, cameras have been around forever. 30 years ago, they're using cameras to detect, parts that weren't in alignment, right?

And cameras were truly that a camera. And you could have some very rudimentary ways to identify if an object was out of place or something like that. But now a camera really, it’s always been a sensor, but now it's truly a sensor. And that's because you're no longer just using it as a picture. You're using it as a true sensor.

You're analysing all the information around it. The facial recognition, age, all those kinds of things that we've seen, the last five or 10 years, you've seen little glimpses of that, and really high processing, but now with edge AI, that's all coming to roost where anybody can put it in because.

The models have really been refined to the point and the compute power, and the cost of the compute power has come down so much that all these things that you had to do, like out in the cloud. Now you can start doing, locally. So really AI is, you always had IOT, IOT, everybody knows that's there, but now IOT has enabled AI to take it to the next level.

That's, where I see, that's, it's just expanding the capabilities on where we're at today.

Ruth: So, it might also fuel adoption because when you just mentioned some of the development of, for example, camera sensors, I always. Like even thinking only 10 years back when some showcases at CBIT where they had really cool ambient assisted living or marketing showcases using sensors and then today you still don't see them in use, right?

So, there has to, there had to be something to stop the adoption of the cool ideas that have already been out there like 10, 20 years ago.

Cliff: Yeah, I honestly believe that the cost of compute power is really the number one driver that coupled with the mass adoption and understanding of how to implement AI.

That, really is one of the biggest challenges is how do you truly understand it? I'm an old hardware guy. So, when Raspberry Pi first came around, I thought, huh, an operating system, that's not assembly. And it was just, it was completely, out of my realm. And Jason Kridner, who, he ran Beagleboard, one of the things that when I was working with him many years back, he said, a lot of the system programmers out there, when you say, just connect this LED, They, they're afraid to do that because they don't know what's going to happen.

And I thought, wow, that's a completely different mindset because me, I just plug stuff together. But you start saying, yeah, exactly, I've got to have drivers and all this stuff. I'm like, where's just my assembly code. And the people on the other side are going, I'm afraid of wires. I'm going, I'm afraid of code. But now. that evolution, the two paths of converged and really to utilize AI, you have to both have the hardware, and you have to have the software, knowledge and background. And that's, really what's key.

Ruth: When you mentioned costs of AI and the evolution of. Things are becoming more quickly.

Do we have to talk about deep sake? I don't know.

Cliff: That one. I just read that over the weekend, and I watched the Stark market take a huge tank and I thought, oh my gosh, that's the, imagine just putting all that money in. And I, obviously I'm not an expert on it, but when you look at that, you think that's how, you learn from other people, every engineer, very few of us are, Developing new things. It's all building off of what other people have built. It's greatest form of flattery, but when it comes to, billions and billions of dollars, it's no longer flattery. it can be considered something else. So yeah, that's, definitely, a big change. That's, on the very bleeding edge, but that will come down. That will come down.

Ruth: Yeah, I think it's interesting that there are now possibilities to make it cheaper or even less energy, hungry. So that is, I think, a really cool step in the right direction. So maybe it also inspires then others to step up their game and follow in those footsteps.

Cliff: Yeah, to be honest, I have, this was. Probably not this last electronica, but the electric electronica. Previously, I was amazed at even what our suppliers were putting on the showcase, because we had, processors that run off a coin, sales that could do inferencing, and I went, oh my gosh. I was like, That's that. It literally, I had to step back, and I was like. I can't believe we're at this place now where you can do AI machine language on a processor that uses a coin cell to operate. Whereas, you know what, five years ago, we're five, 10 years ago, we're talking about sensors and we're like, Oh, they're so power hungry.

And how are we going to do this now? We're doing machine language on, a coin cell power. That's how fast things are moving these days.

Ruth: And what do you see as the biggest barrier to widespread adoption? What are the biggest hurdles for companies?

Cliff: Honestly, I think it's just the, it's just the knowledge.

It's just, how do I actually implement it? The great part is, that it's out there. it's number one, I'll take an analogy that's a little bit different. If you look at IIOT, you think, why isn't IIOT been able to, really take off in the industry? because you've got a lot of factories and heavy industries and stuff like that with production lines, where you're trying to take IOT and you're having the OT try and contact the IT.

And you're using a lot of sensors and things that usually, system level electrical engineers are, dealing with. So IIOT has not really taken off like we thought it does from the enterprise level, but from a smaller level, it's, it really hasn't taken off. AIOT. Which can be, also translated into AI IOT, I think has, a similar stumbling block, but it will go much quicker because the people that have to implement AI and IOT and just AI in general are usually the people, engineers like us that are board level engineers that are doing embedded designs and system programming.

So, we're used to doing research. We're used to getting on the web. So, for me, the biggest hurdle. Is just that knowledge barrier. The great thing is that knowledge barrier has been broken down because of social media, because of, communities, engineering communities and things like that.

And suppliers themselves, suppliers understand that I need to sell this little piece of Silicon in order to sell that Silicon, I've got to build an entire framework, a software ecosystem just to drive the adoption of that one piece of Silicon. So, they have to put a lot of money in. So, what has, what is one of the biggest barriers you're, easily able to leapfrog that and make it no longer a barrier because you have development kits.

You've got communities with people doing projects with all these different models that are out there. When you understand the basics of AI and the basics of machine learning and how it's actually put together. That's the first step. The second one is then just looking to see what other people have done.

So that way you can take your own spin on it. That's where I see the biggest hurdle, but it's not a hurdle that can't be overcome all the information's out there. The biggest problem to your point is. The change, how fast things are evolving and how quickly new models are being developed, it's, that's the hardest part because technology has never moved faster and I've been in this industry quite a long time and I always talk about an engineer's job is horribly difficult these days, I'd hate to be back on the bench because, Way back in the day, you had to figure out how to get the power supply to not, to pass emissions.

So, you had to be an EMC expert. Then wireless came along. You're like, oh my gosh, seriously, I've got to get that going. then it went to LCD displays, modularized power supplies, and now you've got AI. So, if you're, at a smaller to midsize company, you're still the chief cook and bottle washer.

And you've got to be able to handle all this and take AI in.

Ruth: And it's a lot of research and a lot to keep up with what would your suggestion be because it's obviously also a time issue if you have to deliver your projects and you can't be researching and reading up on the recent developments all the time.

Cliff: Yeah, I mean what I do literally I listen to podcasts I was just driving down to Florida not that long ago and I was listening to Some podcast, it turns out was actually just an audio book, like a dummies guide or basics of AI and general and AGI. And it was great. Cause it must've been written by obviously an engineer because it really broke everything down into, the high-level aspects and then looking at the sub levels each way.

So, number one is just spending the time and researching it. And honestly, most engineers that have to develop it. We're in this business because we love technology. We love to learn things. So, the, content is out there. It's just a matter of trying to find it and really, absorbing it. That's where I see the first knowledge or the first thing that you need to do.

Second is just utilize what other people have done. Understand what demos they've put together, like suppliers, like I said, suppliers have all kinds of demos, facial recognition, object detection, vibration control, predictive maintenance. They're all climbing over each other to get those things into people's hands.

So, it's out there, just see what suppliers have done. And then third, literally just get on communities, watch videos, see what other people have done, because that's where we get inspiration from is to see what other people have done and then put our own spin on it.

Ruth: Absolutely that's great advice can you share some use cases that have really inspired you lately some stand out IoT projects or industry use cases that you found really amazing.

Cliff: there's one actually was just talking last night about this you know my mother-in-law is. She's not going to care. She's over 90 years old. So, one of the things, and this actually, somebody brought this up to me at the last electronica, my mother-in-law is over 90 years old. And a lot of times she doesn't pick up her phone and she live alone.

And we're thinking, how do we, find out what's going on there? And. Back in the day, I had like old security cameras set up in my house when my mom was living with us. And I had to do all this crazy esoteric stuff to try and get it, so that I could dial in and see it. And like I said, I've never been a software guy.

And so, it was really difficult. that's, one use case where I think, okay, we need to be able to monitor people, monitor if they fall on stuff like that. So, we actually just put some, the web connected cameras in there. She doesn't really care. So, it's fine for us. We didn't have to have a big argument or anything, but now we can see if she's not picking up.

We can see. Okay, she was walking, through the hallway, an hour ago or something. She's fine. But the best use case that I just, somebody was talking to me about was doing that same type of thing like patient. yeah. Patient or elder monitoring using things like say millimetre wave or something, that you can't really tell how the person's dress, what their hair looks like or anything like that, but you can detect their movements, and you can detect their walking.

You can tech if they fall, things like that. Like when we go through the x rays at, at the airport, you're standing there and you're doing this whole thing, that it's just literally. Checking your outline. I think probably one of the biggest innovations where I go, this is where technology can really come in because a lot of people are like, I don't want cameras in my house.

I don't want people. I don't want my kids watching me wander around. But when you're, I'm not sure the proper term, but when you don't have anything that's built in. Personally identifiable, but it can track your movements. It can track. If you fall, you no longer have to have that little button that does around your neck to call the ambulance, which may have just happened last week, at, at, an in laws house where the police and the fire department came because the button got pressed.

No, you have technology now that can actually just monitor motions. And if they see like a fall or something like that, then, there's an actual real accident. So again, all this technology, we're not really changing a lot of our lives and things. We're just making things better. And I see that's where I, the biggest improvements come is taking technology and just to do things better than how we're doing them today.

To me, that's probably one of the best use cases of something that you think, wow, there's millimetre wave technology out there that people are developing right now for these types of use cases, which is, in the hands of everybody. everybody has access to this, which is just crazy.

Ruth: Yeah, we always, are really quick to talk about industrial use cases or automotive, but healthcare, I think, is such an important and vital issue, especially when, it comes down to who's going to take care where I live. There's a shortage of personnel that can take care of elderly people.

And I'm not sure how it is in the United States, but I'm sure probably not much better because we are all getting older and there's not many people who choose this profession anymore. Is there any, anyway, healthcare. It's such a nice, use case with it, where they get, we get so many ideas and brilliant solutions that actually then also, yeah, really value the, not the patient, but the person that can still live independently at home without having maybe to go to a home or having to have a nurse at their side 24 seven.

Cliff: Yeah, I don't know how much it costs over there, but it's insane over here to have somebody come and live with you. That's why, a lot of people are. They bring their parents in because to go to an assisted living place or anything like that, or it's, literally, it's not affordable for 99 percent of the population.

It's just not viable. So, these types of technologies and these innovations. You look at, the Apple watches and Samsung watches and everything like that, all the healthcare things that you can do now in track, it's just moving it. So that way people are, they're nice buzzwords. Oh, we want to make people's lives more, more enjoyable and everything like that.

It really does come down to how long can you stay in your house and how long can you survive and live on your own and still have the quality of life without having to give up everything. I'd like driving cars. Think about it. If, if my, I wish that, we could get her a Tesla that drives itself everywhere, but, if she could have a car drive her, automatically and not have to take Uber or Lyft that's coming.

It's just a matter of how soon that'll happen that allows somebody to stay at home and not have to have, go have groceries be brought to their house and things like that. and even simple things with a really big impact is the oven still on, or can you turn it off remotely? Things like that also, right?

Ruth: That'd be, really valuable for a lot of people.

Cliff: Every one of us that has an aging parent or has heard the stories of the oven being left on and, that’s off limits now, because that's just can't, can't have that.

Ruth: No, and that was one of the use cases I mentioned earlier, but I saw them at the CeBIT convention here in Europe, really cool, use cases where you had like, a bathroom mirror.

That was basically an iPhone.

It was huge. And it sensed when someone came into the room and then it would remind you of taking your pills. It could recognize your blood pressure. It would do a vital, the basic vitals check. It would sense if the water was still running. Or if somebody had fallen to the floor so that was and many more use cases and that was just yeah really cool things you could do with sensors and then that was even before Alexa’s came along and now you can just put an Alexa in every room and then you can check in with your mom.

Cliff: And that's not that's where the next phase is going to be moving to as well you know all those things that we saw ten fifteen years ago they're now moving towards real adoption. and I, voice control being probably the one that I know will be coming along because, like you said, you've got Alexa's and Google's.

You have these modules set throughout your house and you got to buy all these and have one in every room. And it's great when you can get it to work, and you have full house speakers throughout everywhere. So, I'm wandering around my house. I can listen to Iron Maiden in every room that I go in, I feel like I'm in a really nice place, but eventually we're not going to have to do that.

Everything will have voice control in it. You're washing machine. You're not going to have a Google, in there. You're just going to speak to it because the cost has come down, the accessibility is there and it's just a matter of getting it implemented. Who's going to be first to market with it for real applications.

Ruth: I have the feeling that the development of voice technologies has a little bit slowed down in recent years or is that just my perception

Cliff: I’m honestly surprised that it hasn't been adopted quicker you know that's exactly

because yeah

it's been there and honestly, it's. It's just a different variance of, of vision.

If you go to all the shows and everything, is around object detection, facial recognition. And honestly, when you look at object detection, that's human tracking. It's, packaging it's, understanding if you've got quality issues, so I can see why that one is so far ahead, there's so much work being done because if you think about it, cameras, millimetre wave, infrared, LIDAR.

Everything is around object detection, and I think it's probably getting driven by the automotive industry, right? Not necessarily just automotive, but the automation of any kind of vehicle or machine that moves in an automated manner. That's where I see, the most effort and work. And that's probably why we're seeing it more.

The voice control, it's a nice to have, I'd like to walk up to my washing machine and say, hey, here you go, but I'm not willing to pay 200 more, 250 more to, instead of pressing a few buttons, but to have object detection and facial recognition and all those things, those have really significant.

Meaning in terms of our daily lives, in terms of businesses, being able to save money, do things faster, do things better. That's probably why we're seeing it because the voice control, it's just, pure data from a sensor, So I think that's probably why we're seeing it. That's my take on it at least is, it's a nice to have, but it's not as critical as being able to have your food cart that's automatically going to walk and come through to the restaurant.

That's important

Ruth: in some people's eyes. Any other use cases that you would like to share?

Cliff: I don't know about specific use cases. I'm amazed at the number of use cases that people come up with. And what always, what I really appreciate is if you look at the platforms that are coming out nowadays, I'll use Raspberry Pi for instance.

Raspberry Pi has AI, capabilities. They've got hats, they've got a camera, things like that. And when you take a platform like that, let's say 75 percent of the engineering community, and I think that's probably a conserved estimate. We've all played with Raspberry Pi before, right?

It's something that we've all, messed with. Oh, there you go. That's what I'm saying.

Ruth: Holding it in the camera for the audio only listeners. There you go. That's right.

Cliff: Exactly. the. The democratization of technology. And in this case, AI is really spurred on by the fact that Raspberry Pi has all these capabilities and you have this huge, probably one of the largest developer networks in the world, because you have, community, you've got engineering communities like ours where people come in and they're using Raspberry Pi and Arduino and other types of platforms to say, Hey, Look at what I've built and, people can then take that.

They get the code, they learn from it, and they go, Oh, that's a great idea. I'm going to take my own spin on it. So really, it's not so much about the killer use cases or anything. It's the availability to see what other people have done and have access to that software, that hardware in a low-cost format, in a way that every one of us can understand, you don't have to read up.

Like back in the day, we, we used to have to read a hundred-page, app notes just to figure out how to get the microcontroller to run. You know what I mean? You had to go through every single page. Nowadays you can get on YouTube. You can look at the code, the programming, languages, Pythons out there.

Everything has been simplified so much that it's easy to digest. And you can then jump right in to figure out what you need to, replicate it. To me, that's really, I think probably the biggest having the biggest impact and that will continue to move down that path, especially the hardware developments.

I think SBCs are going to, not just the Raspberry Pi and the Arduinos and the others that are out there, but SBCs in general are going to continue to grow and, Integrate more sensors, look at sensor fusion that to me is probably one of the best, the best advancements that is now enabled due to AI, because when you've got a sensor cube, that's got six different sensors on it.

Yeah, you can, basically build anything that's going to have motion to it. Problem is before; how do you take all those data streams? And that's where suppliers are really doing the, all the semiconductor suppliers and, even sensor companies, they put the time and the effort into, yeah, we've got a lot of data here, but here's how you can actually utilize it.

So that's, to me, those are the enablement of those technologies. It’s what's catapulting the adoption of all these technologies and AI in general.

Ruth: It has become really accessible. As you say, you can just look some stuff up on YouTube, take a Raspberry Pi and off you go.

Cliff: Yeah, exactly. Exactly.

And that's where that, has helped, all the people that come out of a computer science background. You know what I mean? They're now able to jump in. Whereas, the hardware, I think it's honestly hard. I think it was much harder for me as a hardware guy to go deep into the software realm than it is for the computer science and people to come into the hardware realm of things because the hardware's becoming in, we've heard this for years, but hardware is becoming less of a differentiator and it's really all in the software and that's because the hardware is becoming more standardized and mainstream and.

Suppliers and other people are putting together reference designs that literally you can take and put it into production. Assuming you do all the testing of the tolerances and everything like that.

Ruth: So, if an engineer or a developer is completely new and we've touched on some points already, what would you propose should be their first steps?

Or have we covered this enough?

Cliff: I think it goes back to one thing that I learned way, way back when, back when I was at ST, I spent a lot of time with the designers, cause I was a field application engineer working in lighting. And I spent a lot of time with the actual physical IC designers understanding the internals of the IC, how it actually operates, not the black box.

And what I found was that allowed me to understand failure modes in the field, because when you think about a black box, you think, okay, it just failed. Who knows? But if you understand the internal workings, you have a better sense and a better feel for why things are failing. I think AI, it's, it's.

Just the whole reason I've been, listening into that, listening to that audio book is because number one, if you understand the premise, not the premise, but if you understand the basics around AI and machine learning and kind of neural networks, I don't really understand neural network stuff.

And every time they say, yeah, the model works. We have no idea how it learned to its own language. I go, okay, I don't feel as stupid now. Cause you know, I'm like, they don't figure it out. They're the ones that are designing it great. But if you can understand the basics of it and understand how things are operating and how these models are developed, just having that knowledge and that base understanding.

Gives you the comfort and the confidence to then go ahead and start designing it. Because then you can say, okay, now I know what kind of model I'm going to use. And I know what kind of sensors I'm going to use. I'm going to go ahead and start playing with it. So, for me, it's really about understanding the basics and understanding the true technology behind it and how things operate.

If you don't have that long term, you're not going to be able to design a really effective product, because if you don't understand the. To a, to an extent the inner workings of what it is that you're designing with, you're going to run into pitfalls and flaws.

Ruth: Curiosity and, the willingness to learn, right?

Cliff: That's what drives us.

Ruth: That's the secret.

Cliff: Yeah. So, it gets us up out of bed every day. One of the things.

Ruth: Are there any trends coming that, people should watch out for? I'd

Cliff: Say probably the. The one, trend that's not there that we, had an IOT, is security. That's probably the biggest thing because you think about it.

You got people like me that if you got somebody like me that can go and develop something and everything like that, because of everything that's out there, I go, Ooh, yeah, look what I've made. I did this. I'm wearing up my house and I'm doing all kinds of cool

stuff,

but I got it working. Security is the last thing on my mind, but I think security is obviously going to be one of them.

Biggest things that are going to be, is going to continue to trend and not enough people understand it. It's in the news. We all hear about it and there's, trusted platforms and everything like that. But to me, technology is racing and we're all trying to keep up with it, but we got this anchor that's called security on the backend.

And we're going to have to get that figured out because. security is, it can be used for a lot of nefarious purposes and with everything being connected, that's in every country and every individual people, whether or not you're worried about personal freedoms or, liberties and things like that.

I know in the last year, or I'd say last two years, I've probably gotten three notices in the U S you get notices from companies when they're being sued for, some kind of a breach or stolen identity type things. I've gotten three of them in the last couple of years, just around fingerprint applications where I was not notified or somebody got access or something like that.

Then that's just fingerprints. now we're doing facial.

Ruth: They're trying to make things secure.

Cliff: Exactly. Already.

Ruth: But then, okay.

Cliff: Exactly. But now it's going to be facial recognition. It's going to be retinal scans. It's going to be everything about us. So, I think that's going to continue to be the trend because companies have woken up, they don't want their video cameras being part of a, a DDS, attack, it's just not, you don't ever want that.

So, to me, that's probably the one area that's going to be continued to trend. To be the number one leading trend that people are going to have to do. It's not fun. It's not sexy. Sorry for you guys that work in security, but it's, it's a necessary evil that we have to, we're going to have to continue to, focus on.

Ruth: No. And that's, of course, there's also something hard to balance if you want to be. Ahead of the pack and being very innovative and, doing the cool stuff, but at the same time you run a business, and you have to make sure that you protect your customer and your clients. That's really tough to handle.

Cliff: Exactly. And like I said, it's not a glorious, thing. It's like the old, in the old days, the guy or the lady that had to do the EMC testing, nobody likes you cause you always end up having to make people go back and redo things. because if I got noise on the line. It is not going to pass, and security is going to be the same way.

Ruth: Is there anything that I have not asked you that you wish I had asked you?

Cliff: Other than who my favourite Marvel character is? No, I don't think so. That's, I think we're, I think we've covered most of the, most of the topics.

Ruth: Now you have to tell it who is your favourite Marvel character.

Cliff: It'd have to be Iron Man, of course.

You have to be iron man. He's a little cocky, but he's got the, he's got the goods to back it up. So, I give him kudos and you.

Ruth: and franchise wise or movie wise, he's the original, isn't he?

Cliff: Yeah, that's right. The original, the innovator. And how about you yours?

Ruth: I have to go with black widow.

Cliff: That's, a fair call.

Ruth: Although she doesn't have. Superpowers, but

Cliff: hey, neither does Tony, but she holds her own. She's definitely she belongs in there. She's also

Ruth: but I do also like Jarvis, I must say, and I would really love to have him or one of his copies.

Cliff: Yeah, there's I just instead of saying, hey, Google or Alexa or anything, I just want to say Jarvis, why can't we get that?

I was going to put a petition out years ago and I thought if I, on my bucket list before I die, if I could just say Jarvis, that would just give me that much more happiness in life. that's really one of the, it's one of the few things on my bucket list is to be able to call my home assistant Jarvis.

Ruth: Everything complete and instead of scrolling the mouse wheel I want to move stuff with my hands

and then

dive deeper into the information and pull it out and zoom in and.

Cliff: Yeah, exactly problem is we're all engineers and we'd like the big screen but I’m like yeah not until I can get it for free or I can get it for a hundred bucks.

Ruth: As a test example, please.

Cliff: Yeah, exactly. It'll happen. It'll happen soon enough.

Ruth: If you had to put together a soundtrack for this episode and I've read up on your bio, you are also a big music lover. What song would you put on our soundtrack for this episode?

Cliff: Honestly, we're going back to, we're going back to Marvel.

So, I'd have to say, the, what was it? ACDC, when Tony Stark came in, at, the Thunderstruck, it'd have to be Thunderstruck.

Ruth: Cool. Very nice.

Cliff: Let's see if we can make that happen.

Ruth: I think so. Yeah. And I will also always, we now actually also have a, we talk IOT playlist on Spotify and on YouTube and I put, my guests’ songs on it each episode.

So, then you can listen to. To the whole compilation at one point.

Cliff: Excellent.

Ruth: Thank you so much, Cliff. This has been an amazing talk and a really interesting show. I hope you enjoyed yourself as well and thank you so much for being here.

Cliff: Yeah. Wasn't even work, so I appreciated it. It was great.

Ruth: Terrific.

Cliff: Thank you.