window.dataLayer = window.dataLayer || []; function gtag(){dataLayer.push(arguments);} gtag('js', new Date()); gtag('config', 'UA-104066415-1');
Listen on iTunes

Mister Beacon Episode #135

Ultra Wideband Technologies for Car Factories

September 14, 2021

This episode we’re talking with Andy Ward, who is the Co-Founder and CTO of Ubisense, about various technologies from ultra-wideband, to Bluetooth, machine vision, and many more, which his company utilizes within car manufacturing factories to automate the assembly line process.

Listen as we discuss these various technologies, old and new, and how in combination they have the chance to change the manufacturing process from top to bottom.

Transcript

  • Steve Statler 00:03

    This podcast is sponsored by Wiliot, Intelligence for Everyday Things powered by IoT Pixels. Andy, thanks so much for coming on the show. It's a real pleasure having you on. I've enjoyed the times, we've got a chance to talk over the years, and I can't think of a better person to talk about how you choose the right technology for the right job, specifically between ultra wideband and, and Bluetooth angle of arrival. Although for people that want to stay for the second half of the show, they'll hear some really interesting stories about how you've got into infrared and ultra wideband, which is another set of choices. And maybe we'll get onto that later. But first of all, welcome, and thanks for coming on. Thanks for having me. What is that interesting array of objects just over your right shoulder.

    Andy Ward 01:04

    So this is, you know, we do up sense, we may care, location aware systems for industrial customers, and, you know, about 5% of the world's cars are built using our technology now. And we sell ultra wideband systems, we sell third party location systems, including ble AI systems from other people to these manufacturers. And then systems work in terms of software and installation management that goes with it. And as part of that, particularly for these customers, you have to make sure that he keeps working, you know, they have extremely high reliability requirements they have, you know, you're allowed basically zero downtime, and it's got to be the right, right answer every single time. And so a lot of big systems engineering has to go into making that work. So this here is part of our large scale, one of our large scale sensor test rigs, where basically we have about him is he gonna be 5060, something like that sensors behind, they're all working in concert with a cloud server, and so on simulating some of the kind of traffic and traffic patterns that we get in large scale industrial systems. So we check it out here before we roll it out to mission critical sites around the world. So is this part of product development? Or is it part of the QA process for every beacon that you ship? Oh, it's this is this is actually more to do with the system software side of things. So of course, when we ship sensors, and when we ship beacons, we do have QA processes, of course we do. This is actually more of the software, the system control software and application control software, you know, you have to basically simulate large amounts of data. And we do that actually write in the sensors themselves, so that we can simulate all the paths up through the the software into the application layer to make sure it really is going to work as reliably as we think it is. So just something we've learned over the years, you have, you can't take any chances with this, because you know, Nate nature out there is much more cunning than than you could ever be in terms of doing it. So the more you can try it in a in a representative kind of way, the better your your system reliability will be. So for those of you that are listening, which may be a large proportion of this, this is kind of an array of

    Steve Statler 03:42

    devices, that are maybe the size of the palm of your hand, roughly a paperback book size. Yeah. And so those are those ultra wideband beacons.

    Andy Ward 03:55

    Now, these are actually the these are actually ultra wideband anchors or sensors. So these, okay, that are placed around the environment and pick up ultra wideband signals from from beacons. And then from there, you gather the data from those calculate positions, but then they go up through further levels of, of software. So, you know, you take computed XYZ positions, and then those are turned by software into things that businesses really care about. So you know, nobody, nobody actually cares that much about XYZ coordinates while they care about his business relevant events. So this is an industrial moment, say they care about business relevant events, like this tool is now working on that car, or this car is now in this process space. You know, those are things that the software then works out. And then on top of that, you have further layers of software that do things. So you know, when this tool is next to this car, program it using these particular parameters and that's something in the manufacturing execution system or the, the control system for the factory. So, you know, to make all of that work and all that work reliably, we do an awful lot of testing, to check that, you know, when we take our software products and deploy it in factories that the factory is going to keep working, and it has to have you stop something like a modern car factory, then it's something like 300 euros a second is the cost of the downtime. So it's absolutely essential when you deploy these systems that they work, and they work first time, and they keep working for years. So

    Steve Statler 05:38

    and how are those things powered and connected?

    Andy Ward 05:42

    The anchors, generally, they're connected via Power over Ethernet. So, you know, I think, a lot of you know, pretty much all of the location technologies that people use indoors these days, you know, from, from a XYZ accurate positioning kind of sense, you know, they tend to have a fixed infrastructure, that fixed infrastructure usually has to have power. So it's, it's, you know, normally something where you run a Power over Ethernet cable to it, and that supplies, both the power and the data, and you can't normally do better than than one, one cable running to these things that does both things. I think, you know, in the future, that might be something that changes when there are, you know, there are new technologies coming out, like 5g that allow, you know, reliable wireless interconnection of that kind of infrastructure, you won't get rid of that single power cable, it's still a, you know, you got to still provide the power one way or another. But it might be that that power comes over a bust network, you know, some like AC mains, or maybe a high voltage DC bus, where it's actually a lot easier and cheaper to put in a new spring than it is to, you know, hook up a POE line back to a switch and a switch cabinet. And

    Steve Statler 07:05

    what, it really is easy, I would have thought POE would be the easiest because it's not power a power cable, it's a data cable that just happens to have a bit of power there.

    Andy Ward 07:15

    You can't kill anybody with it. Right? Yeah, I, you know, it's funny, I think the issue isn't one so much of physical ease of running it, it's more the cost of maintenance in these factories, in terms of making sure that your switches are appropriately provisioned. And that the switches are, you know, protected against everything else in the factory and so on. So, mains is pretty, it's pretty bulletproof, right? And there's not a lot of thinking that you have to do when you add a new spare remains. Whereas, you know, network provisioning and network management and so on are something that actually have a big hidden costs when you go and put those in. Oh, interesting.

    Steve Statler 07:57

    Yeah. Cuz I said of it was the most expensive thing to do is to pay an electrician to start laying infrastructure. But I guess what you're saying is the data team actually get paid even better than the electricians?

    Andy Ward 08:11

    I think that I think that's probably true. Yeah, yeah. Oh, so

    Steve Statler 08:14

    how many of those Do you need in in a factory to track? Everything? And what are you typically tracking in a car factory? Presumably cars and bits of cars, but

    Andy Ward 08:27

    Well, yeah, so I mean, I think the answer to the How many do you need is a little bit of a open ended question. Right? It could be, you know, we have we have factories, whether there are four of these things, or, you know, three of these things. And there was a factory where there's that 1500 of them. Right? And, you know, it depends on the area of the factory in the kind of applications that you're looking at, and so on. So, you know, a good a good application example, is that control of tools. So, you know, one of the things that's really interesting for you, we sense and I think where we were, we have some experience and expertise is in using location information to directly control industrial processes. So somebody like BMW or Daimler, their factory these days is completely different from the factory they had 20 years ago. So when when Henry Ford came up with his idea of that, that you know, the Model T for production line, you have this genius idea with mass production, which is you make the same car every single time. And every time you do something at a particular station, on this moving assembly line, so the cars move past you as they're being assembled. And you as a worker, do the same thing every time to every car that goes past and that was transformational in the way that it you know, accelerated the car industry and reduced costs. The problem is that every time That assembly line has to be the same, because you don't get the opportunity to chop and change between different options and models and variants. And that doesn't sit well with, you know, today's today's Moore's which are that, you know, I want my car, which is different from everybody else's. And so what people like BMW and diamond have done is they've put, they've made production lines where every single car can be different. So you know, my car that I want starts being made on the production line, and then the one after it will be completely different than the one after that will be completely different as well. So we may actually have four or five different models on a production line, we may actually have, you know, millions of variants, so each car is different. And that means that when you do something to a car, as it passes you on this moving production line is a worker, like us a, you know, a high end screwdriver, essentially, to to put the airbag in, for example, you you actually have to set up that tool to do the right thing for the car in front of you. Because if you use the wrong talk or use the wrong number of turns or something like that, then it can basically damage the the frame and or maybe, you know, like the airbag doesn't go in, right. And essentially, you know, you have a real problem there. So what those customers now do is that they track all of the tools on the assembly line using tracking technology. And they track all the cars on the assembly lines, it moves down the line. And they use software to essentially look at the location streams coming in. And when a particular tool is being used on a particular car, instantly, the control software recognizes that and sets up the tool to do the right thing for that particular car at that particular point on the assembly line. So from the point of view of the worker, they pick up the tool, they use it on a car, and magically, it's set up in exactly the right way without them having to do anything. And so, essentially, that's how a modern high end car production facility works. And so to do to do that, you might have, you know, the biggest factories, we've got, you know, maybe four or 500 meters square, so, you know, sort of no third of a mile on the side, and, you know, may use 1000 sensors to do that. So it's, you know, they're quite big systems, and they have to keep working all the time. So you can be working one moment on the hood of one car, and then turn around and you know, a fraction of a second later, 200 milliseconds later, you can be working on the tailgate of the next car on the assembly line. And in that time, the system has to have track the tool recognized, it's moved from one car to another, reprogrammed it so it's now doing the right thing. And, you know, done that with with, you know, an error rate, which is kind of probably less than one part per million is what's permissible these days. So well, you know, it's a very high end kind of tracking system. And for that, you know, we typically use ultra wideband location because of the high tracking reliability that's necessary. You know, it's an A high tracking accuracy that's required in in what's quite a challenging environment with a lot of metal and reflections. But obviously, that's, that's the right technology, the right sensor for that particular job. And it might be complete overkill for a different kind of location tracking challenge, or is not enough. I mean, there are other technologies, which are even more accurate than ultra wideband. So you always have to choose the right thing for the right job.

    Steve Statler 13:36

    Yeah, it's a toolkit. And there's always pros, cons cost, and infrastructure and battery life and robustness and all these other things. So, how accurate Do you need to be to know the difference between the end of the back end of one car and the front end of the one that's coming quickly? behind

    Andy Ward 14:04

    it? That's a good question. I mean, they're, they're separated by about half a meter or so. So you know, at some level, you have to be kind of, you know, accurate to that kind of level, that the biggest challenge actually is not so much the, you know, instantaneous accuracy in a nice, clean environment, it's actually dealing with the fact that, you know, as you're doing this, the tools are sometimes used in the wheel wells, or, you know, under the car or in environments that are, you know, not not ideal at all for propagation. You know, I think I think people have their mobile phones given them a false sense of security about how radio signals propagate around the environment, you know, that, you know, they don't ever really see the fact that you know, radio signals are blocked by all sorts of things, people I mean, I you know, people are fantastic radio absorbers. wet, salty, you know, bodies are fantastic at just soaking up radio waves, as are bits of car, you know metal, metal parts of car or metal parts of production transport systems or something like that. So, in those environments, you know, it, actually you very rarely get really good propagation from beacons on tools or on cars into infrastructure that you place around the, around the production line, you know, you're always being challenged by reflections, and obstructions, and so on. So even if even if your technology works at 50 centimeters in a nice, clean environment, the question is more, what's the level of reliability and robustness? You get when you're you're dumped into a hellish environment? And is it good enough to do it there with the levels of reliability that then people need? And, you know, it's a it's a challenging problem. So

    Steve Statler 15:54

    the presumably even ultra wideband is not going to propagate through sheetmetal. Or, or someone's body, is it? No, no,

    Andy Ward 16:04

    you're right. And I think that the, you know, the, the important, the important thing there is that actually, you have to put up enough infrastructure that at some point, you have a, you know, a few near lines of sight, right. So, you know, if you go from line of sight and near line of sight, your accuracy of your system degrades, I mean, just just going to right, if you go from near line of sight to no line of sight, or blocked, you know, then, you know, things get a lot worse a lot, a lot more quickly. So, you know, a lot of these car production lines have to have enough sensors that you're getting the the basic number of line of sight or near line of sight paths, even in those, you know, really challenging kind of situations. And then it's down to, to how much information do you squeeze out of every line of sight or near line of sight path that you're getting? So if you're, you know, and this comes back down to location system architecture, really, right, so if you, if your location system is a pure tdoa location system, you'll need four

    Steve Statler 17:14

    sets time difference of arrival,

    Andy Ward 17:17

    arrival, sorry, yes. So we've got a pure time difference of arrival system like GPS is, or like, you know, there are many radio time difference arrival systems, then essentially, you need four of those pretty clean or nearly clean paths to get your 3d location with any accuracy. If you if you do two way ranging, so that's normally called tra, or time of arrival, where you're sending a signal out from a tag to an anchor, and it comes back again, and you measure that the two way trip, you might need three, right. So that's kind of a bit better and a bit more robust in those really challenging circumstances. If you use a away, actually, you only need to, you know, that's an even more robust technology. And we actually use a combination of eo a plus tdoa. So that actually gives us a lot of information from any path that happens to kind of make it through the the maze of the maze of obstructions that you're placed around.

    Steve Statler 18:18

    Now, that's interesting. So you're, and I always call this if it's on an asset that's moving around, I tend to call it a tag and I try to call beacons of things that don't move. But but the complicated thing is that tags tend to use beaconing radio protocols, which is really confuses everyone. But anyway, let's get back to the point that this mixing of AI and time of arrival so you so you're using both and ultra wideband with ultra wideband is that the the the time of arrival, that your time difference?

    Andy Ward 19:04

    We make the ultra wideband does both so you can measure an ao a on a ultra wideband signal in the same way that you do on say, a bl okay, right. So right that might have it's very, very similar. It's a bit easier because of the shorter signals to measure that tdoa on a ultra wideband signal than a ble signal. So, you know, we have the opportunity with ultra wideband to do both of those techniques at the same time. And they complement each other quite well. But essentially, what you're trying to do is, is you know, you've got a direct path, squeeze as much information out of it as you possibly can, because you may not get many more of them, right. So, so from a robustness point of view. It's a very robust technology, but it's probably you know, it is it is a bit more expensive than ble. So, you know, if you look at the kind of applications that you have in car plants, then ble Might be a great application for logistic store areas or something like that, where, you know, you either you don't need quite the accuracy or you know, all the quite the robustness at the same time, it might be a great application for those, whereas somewhere on the production line, you know, it's a very nasty radio environment, and one needs to, to squeeze out as much information as possible with the highest bandwidth signal. So, you know, in my view, when we put together location systems, it's always horses for courses use the the best technology for the particular problem that you're looking at. And, you know, I although we make ultra wideband sensors, you know, there are applications where ultra wideband just isn't good enough. So, you know, we work with car manufacturers who want to do what's called bolt level accuracy, where you, you know, if you if, let's say, you've got a good example is a helicopter is that is the rotor blade assembly for a helicopter, you have lots and lots of bolts, and they have to be bolted down in precisely the right sequence. Now, kind of imagine changing the wheel on your car, the tire on your car, you know, you're supposed to kind of tighten the bolts on the on the wheel, in a kind of, you know, if you've, if you tighten the one here first, then you go across the wheel to the next one, the next one. So there's a particular way in which it's, you know, we're doing it well, for a helicopter, plate assembly, it's like that, but there's 50 bolts, and you know, the consequences of getting it wrong are really bad. And all of those bolts look alike. So there, you need to basically be able to tell, within a couple of centimeters, where the tool head is, as you're assembling this thing to make sure that the assembly process has happened exactly right. And that's beyond ble, or ultra wideband, in fact, probably the only way of doing it is optical. So you know, for those kinds of applications, you know, we we suggest optical techniques, because it's the only way of really being sure about about what you're doing. So there's a continuum of, of location technologies, each of which is applied clickable to different industrial kind of applications,

    Steve Statler 22:15

    and the optical is that machine vision, or is it visual light communication? Where you're, what what?

    Andy Ward 22:26

    Yeah, that's a good question, we work with a company called AR t out of Germany and AR t have built a tool mounted camera. So it's kind of like a little camera that looks directly at where the tool is, is being used on the assembly, and it looks at the area around the bolt to, to figure out, you know, what, what, which bolt you're looking at. So they actually originally did it for things like engine assembly in, in car production lines, right, because, you know, if you imagine kind of trying to work on a, an engine deepened an engine that's in the kind of engine bay of a car, and as a, you know, metal hood above you, you know, it's a really awful scenario for anything that's kind of outside the car, trying to detect signals that are coming from a beacon or something like that in No, you can't look into the car, you can't get any signals out, the best you can do is actually look at what's happening inside the bonnet, or inside inside the hood of the car, you know, with the with a tool that's got a little camera on it, and the camera can see the area around the bolt, and it can see exactly where you are with respect to other bolts and so on. And actually with a bit of machine learning and, you know, Ai, they can tell exactly which bolt it is that you're looking at. And so you can then know that with some quite interesting combinations of technologies because this AR t vision based technology is quite capable of telling you within an engine block, you know, which which individual bolt it's looking at. But one thing it doesn't know is which car you're looking at. So every car, the engine block looks identical to the system. But you can combine that with the tracking from ultra wideband or with you know, ble or something like that, you know, on a production line, we we combined it with ultra wideband because the ultra wideband knows exactly which car that is working on right now. And the AR t vision system knows exactly which bolt on some car it's it's working on right now. And if you combine those two things, you can essentially work out exactly which bolt you're working on on any car in the factory down to the individual bolt. And so you can do quality control processes and understand exactly what's happening at a much much deeper level and you wherever possible, ever able to do before. But that's only possible because using a combination of different technologies, each of which have some strengths and some weaknesses but if you can bind them in the right way, you can get a combination that's strong in both both regard.

    Steve Statler 25:05

    And I guess the machine vision product is not gonna have the kind of the serialization, you have a, like a Bluetooth tag on a on an asset, then that can be indexing into some unique ID. And that's kind of Another limitation. Because I think, you know, for those people in the competitive mindset, the question is, well, what's the best technology? And of course, the answer is, well, is a hammer better than a saw? Now, it's like different things for different jobs. Any other limitations of that machine vision that you see that preclude it from being used for everything?

    Andy Ward 25:47

    Well, I mean, I think it's, it is, it is down to the the identification, right? It's, it's not great identification, um, you can always arrange for it to be good, I've, you know, been a big barcode on something, and maybe that's a good thing. You know, but the, you know, it's, it really, genuinely is a line of sight technology, right. So, that's not always ideal. And, you know, any radio technology that you will get some diffraction round things, and, you know, you'll get some penetration through certain materials. So it's, in my, in my experience, optical technologies are a bit harder, you know, both in terms of difficulty, and also in terms of, you know, being black or white, right, it's like, either it works, or it doesn't, right. Whereas rating technologies are softer, they, they degrade, and they may degree a quick, but it's not a complete, kind of instantaneous cut off. So that, you know, I think machine vision is going to be an interesting technology, and people are going to do some, you know, I've seen fantastic performance from things like slam and so on, where people are now tracking forklifts, where essentially, you put cameras on the forklifts, and you drive the forklift around, and it's sensing, you know, everything as it goes past. And you can see exactly where you are. And you can map it out at the same time. So some technologies that people are already using things like the Oculus, headsets, and so on, you know, in a bigger scale, are being used all over industry, and I'm pretty sure that there's going to be a lot of consumer, drone devices, and that kind of stuff, where that that that stuff is really going to be very mainstream in the next next few years, robot vacuum cleaners, that kind

    Steve Statler 27:35

    of stuff. So remind me what slamm stands for

    Andy Ward 27:39

    simultaneous localization and mapping. So the idea is that if you've got cameras on a device that are looking out, then, you know, you can use that combined with inertial technologies to essentially do what we do as people, you know, we, we know where we are in buildings. And the way we do that is by looking at where we are, and combining that with the information that's coming from our ears. And, you know, we build up a map of exactly where we are and what's around us and how we're moving through the space. And that's what slam does in a, in a computing way. And I think that's something that that has really taken off in the last, you know, 1010 years, I'd say, you know, the technologies to do that have become widely available. You know, one of the things actually, that is a bit of an aside this but one of the when I was doing my PhD in location tracking technologies, there was one paper that was just completely on another plane that that I came across, and it was from a guy called Gary Bishop at the University of North Carolina, who is you know, I met him once bonafide genius, I'm sure. But he built he wrote a paper in about 1982, I think it was about a thing called a self tracker, which was a cube of cameras, you know, each looking in different directions, and it had some accelerometers. And this guy was basically 30 years ahead of his time, right? So you probably couldn't have built it in 1980. Or maybe the military didn't, never told anybody but you know, now is reality and that's an interesting technology for the future.

    Steve Statler 29:20

    Well, I love your metaphor or tie back to human beings and what we do but of course, human beings, augment our visual senses with location technology on phones so we can see where we are we roughly recognize were in the middle of the Yorkshire Dales, but we don't know exactly where else we are. So we use GPS and all that stuff on our phones. So I want to just, you know, this is fascinating comparing the different technologies that are in the toolbox, just going back to the AAPC because we kind of build this as mainly Comparing Bluetooth and ultra wideband but it's so much more do those anchors that you have over your shoulder that? Yes, it's like a paperback. It's sort of like the old Michelin Guide, because they're a bit thicker than your average spy thriller that you might buy at wh Smiths bookstore in England. So do those have Ayoade, those have multiple antennas, or is that something else that you need in order to do a no so that

    Andy Ward 30:30

    so inside those is a kind of an array of antennas, and that's how we know we look at the very, very slight time differences of arrival of the signal, or the phase of the signal each of those antennas. And that's what allows us to get a two axis elevation and Asmath. eo a, and so, at the same time, we can also tell exactly when the signal kind of came in. And that gives us the tdoa, or the timing from A to A to so you know that Yeah, they're kind of about the size of a book. I mean, you know, obviously, our our focus is in industrial applications. And, you know, these are by no means that the the ugliest thing in the factory. So, you know, I think sizes, the size is perfectly acceptable for that kind of thing is that it's the tanks that have to be small, you know, you have to, you have to make small tags that are, you know, have long battery lifetimes. And, you know, particularly for industrial applications, it's got to be pretty hands off, because you don't get the opportunity to recharge everything in the same way that you do with, you know, many consumer applications. So, you know, every, every night, we're all programmed now to plug our phone in to get recharged, right. And so a battery lifetime of a day or two is the sweet spot for fairphone. But for a tag that's used in industry, and this is true of Bluetooth as well as you do be. Now the sweet spot is probably about five years. If you can make something last for five years, then the likelihood is that the process in the factory will have changed on that kind of timescale. And so that that's kind of what they're looking for,

    Steve Statler 32:09

    I think. So in trying to position the different technologies, I've always thought of you wb as having more power consumption at the tag level less battery life, are you saying that that's changed that it's at parity with bayarena?

    Andy Ward 32:27

    Depends, I'm not sure that all ultra wideband is the same if I know it is not the same, right. So So our the technology that we use, which is short pulse, and low pulse repetition rate is on a par with ble, in fact, we actually just launched a, a tag, which is a dual ultra wideband and ble tag, so it does both of them and has a, you know, the same, you know, same sorts of battery lifetime is a pure ble tag or a pure AWB tag, there are other tech, there are other kinds of ultra wideband which are higher power, where they use many, many more pulses. And, you know, they got a lot more circuitry running really, really high rates. But But you know, I personally don't like those so much for industrial applications, because they just a lot higher pass. So so I think that the BLE approach, and the approach we take is probably right for industrial applications, because you need to have something that doesn't require a lot of maintenance, even if you're driving it quite hard.


    Steve Statler 33:39

    And what about the time dimension here? Because that's yet another factor. So how often do I need to know, changes in location does to the two technologies bench out the same as well, if I want to be broadcasting 20 times a second or something like that?

    Andy Ward 33:56

    Yeah, yeah, I think that they're very similar. I mean, the, the ultra wideband stuff is is ultra wideband, again, is probably has probably the edge there because it's possible even to transmit two packets simultaneously, and have the Miss, because the pulses in each are so short, that as long as they as long as they don't act, the pulses don't actually overlap, which is incredibly unlikely. And you can actually transmit and receive multiple packets in, you know, that completely overlap, which is harder to do with a technology like ble, but again, you know, I think, I think the fact that both of them kind of score equally on on the kind of or, you know, similarly, on the on the things the metrics that you're talking about is really a reflection of the fact that, you know, companies like UB sense and so on, you know, kind of optimize their technologies to meet a particular market segment. So, you know, we've optimized it because we know that really, if we don't get five back five years back should I've done then, you know, industrial manufacturers, not maybe that interested in it? And that's true for ble vendors as well. So what's the what's the things that were we, we you we sent don't do so, so much on? You know, because we've made that trade off? Well, you know, for us, you know, our tags don't transmit megabytes of data per second, it's just not an interesting thing for the customers that we do. So our ultra wideband is optimized towards, you know, long battery lifetime. Many, many tags in the same space, low tag cost, all that kind of stuff. But it isn't a data is not a general purpose data, link. It's not


    Steve Statler 35:38

    always so your tags aren't syncing, they're just tagging a location or,


    Andy Ward 35:44

    well, they sense they sense things like movement, and they send things like temperature and so on, but they're not. Yeah, a good example would be something I think like a, like the tools that we use in assembly operations, or our customers using assembly operations. You know, every time you do an assembly operation on a car, not only do you tell it all what to do, but the tool actually records all of the torque data that's that it's measuring, as it does the screwdriver operation. Because you can tell when you've got a strict thread or when you've got a nut that's not down correctly, and it shifts all of that back. And that might be Yes, I'm no 1050 100k bytes of data. And so there are applications where you might want to ship that kind of amount of data that but our view would be, that's not really a location systems job. That's something where the tool probably has its own Wi Fi, or 5g or whatever data link. And that's the appropriate way of getting that data back again.

    Steve Statler 36:47

    So Andy, would you say that you're a musical person or not?

    Andy Ward 36:53

    Yeah, I'd say I'd say I like my music. Absolutely. Yes. Excellent. I generate a lot myself by listening to it.

    Steve Statler 37:02

    I'm the same. I'm the same. I know my limitations. So if you had to choose three songs that had some meaning to you, what would they be?

    Andy Ward 37:15

    Um, so I think the first one would be Wuthering Heights by Kate Bush. And it's just because it's a classic. It's a classic tissue, but it's it's it's the first thing I really remember from pop culture, because I think a lot of people have you know, the story about hiding behind the sofa when doctor who is on because there's some scary monster. Well, the first thing I remember from pop culture is hiding behind the sofa because Kate Bush was on singing Wuthering Heights cuz I don't know whether you've ever seen the video, but it's a pretty ethereal kind of far out video. And it's a five year old that terrifies me. But it's a great song. And I you know, obviously, I've come to appreciate

    Steve Statler 37:55

    it's tremendous brings back loads of memories for me as well. And if anyone hasn't seen it, they should. And at the time, it was just completely different to anything else anyone has ever seen. I think, what an amazing artist she? She is. Okay, fantastic. Great choice. Number two.

    Andy Ward 38:14

    So number two are the, I think the sound of silence by Simon and Garfunkel because for me, it's, I mean, I could pick it almost anything they did, but you know, that's something that takes me back to driving long distances and the you know, in the car with my parents, because they had that, that tape on, we listen to all the time. But again, it's funny how, you know, sound and, you know, those kind of things can can bring you back to a certain place. So that's definitely, you know, driving from one end of England to another kind of,

    Steve Statler 38:48

    where would you drive? What are these drives, drives on holiday or


    Andy Ward 38:53

    just to see, just to see grandparents and that kind of thing. So you know, but, you know, all those kind of things, but the sound the silences is the one that I kind of always remember is slightly haunting again, to be like, okay, bushwalk

    Steve Statler 39:07

    Yeah, very good. Yeah. I mean, Paul, Simon's music really endures. I ever it seems like every time I go online, you can call me out is like at the top of my YouTube recommendations, and it's actually stands up. Great. And number three,

    Andy Ward 39:25

    and number three would be superstition by Stevie Wonder. So, you know, again, it's all kind of tied to a place for me I want in my spare time, I do paragliding and there was a trip I went on to Morocco where the next day I was planning to do a certain maneuver where you collapse your parachute and you know, just as a practice, emergency drills kind of thing. So there's a little bit of me which was like, apprehensive about doing this and so I couldn't really sleep but the rest of the Ruth decided that they were going to party all night. And somebody in that group decided they were going to stick superstition on repeat. So I was like, I was obviously not getting a lot of sleep anyway. And after that, all I could hear was every single note and superstition, but fortunately, it's a good song as a great song, Stevie Wonder is the best. So, you know, it's a fantastic song, I'd rather have that as the thing stuck in my head in that circumstance than anything else.

    Steve Statler 40:26

    So is the paragliding the one where you run down the hill with a rectangular parachute and you glide around or different

    Andy Ward 40:34

    sides? I mean, they kind of they, they're pretty sleek. Now they kind of crescent shaped kind of things? And yeah, you, you sort of run down the hill for a bit. And then often, you know, you're with the birds. So yeah, they

    Steve Statler 40:47

    do that here in San Diego, in La Jolla. And it's spectacular. You go over Torrey Pines golf course, and people are chipping shots underneath your toes. And I'm terrified of heights, but I did it for my 50th. And I'm like, Oh, this is a scarier tour is wonderful. And then the next weekend, there was news that someone died doing it, they they they hit a cliff or something down there. So it's dangerous.

    Andy Ward 41:16

    It is you know, it's a it's a thing where you have to, you know, it's like any form of aviation, you have to treat the treat it seriously right? Because it's, you know, it's not your natural environment, you've always got to remember that.

    Steve Statler 41:29

    So I was wondering whether you'd have any music that conjured up your days in Cambridge, because having been an alumnus of Hatfield Polytechnic, I always sort of looked at that. And, and I had a girlfriend whose his dad went to Cambridge. So we used to kind of occasionally go to these black tie dinners, and there's a lot of really good wine and food that was served. Was that was that a lot of fun? Was there any music associated with that? Or was it just intellectual activities, cerebral time in your life?

    Andy Ward 42:07

    I think, you know, the sort of more classical things, I think, you know, there's a lot of choral music in Cambridge, because it's obviously based around the colleges and colleges are essentially based around the church. And so you know, it probably has more churches per square meter than anywhere in the world. So there's a lot of choral music. And, you know, it always comes into its own at Christmas, right, because then they do a lot of choral music for Christmas, and so on, and let you know, that's always a nice, it's always a nice place to be there. And there's always a lot of that, that kind of stuff. So yeah, it was gonna pick one I'd say, the shepherds farewell by Berlioz if you've never heard that it's a beautiful piece of

    Steve Statler 42:52

    music. I was reading a book by David Byrne of the talking heads. It's fascinating. It's about like lots of different things. It's a biography. It's a treatise on a theory he has on music. And it's also a guide to the music business. And the theory he has on the music on music is that the sound is a function of the place. And so you can never have rock music in a cathedral. It would just drive you crazy, all the echoes, but the choral music is specifically tuned to the the the the echoes and kind of surviving that whereas talking heads and Blondie was the perfect music for cbgbs where the kind of tiny space lots of noise have to kind of get over the bar all this anyway, that's by way of nothing. But you just kind of clued me into that. So what was what was it like going to Cambridge? Because you did. I never quite understand what these things mean. But you got a Master's, which I think is what everyone gets. But you did a PhD there as well.

    Andy Ward 43:59

    Yeah, I did my undergraduate degrees in Cambridge. And as part of that, I actually was fortunate enough to work with a research lab called o RL, which was sponsored by Olivetti was, you know, kind of slightly peculiar thing, and it's time, but I guess now, it would be kind of viewed as an incubator. I mean, that's kind of what it was. It was a way of nurturing startup companies generating some papers and some publicity for all of us at the same time. And they have a thing called the active batch. So I don't know whether you've done a a, I don't know whether you've done a show on the prehistory of beacons. But the active badge was a thing that they built in about 19. I'm gonna say 1988, something like that. And it was an infrared location system. So it was a little tag that you took around with you and it transmitted infrared to a network of sensors around Building and he could tell you which room in the building you are in. So it was a kind of room scale, location system. And then there is still systems like that today. But as an undergraduate, I actually got access to this and use it for my project. So I kind of got into this idea of, well, you know, you could you put a map up, and you can show where people are on the map. And that's not so good. You know, that sounds interesting. And so when I finished my undergraduate degree, and I worked in industry for a bit, I went back and did a PhD on on that kind

    Steve Statler 45:31

    of, well, Andy, I could talk to you all day, this has been fascinating. Thanks so much for coming on the show, I really appreciate it. And I hope people stay on and listen a bit more to hear a bit about the genesis of your company and how you got into this space. But thanks so much for giving us this incredible fascinating guide of the the different technologies that are out there to help people with rtls

    Andy Ward 45:57

    Thanks for the opportunity in the building. But if you want to control systems, and actually make them do something useful and different, so you know, I walk up to a computer, and instantly it's my computer and all my stuff appears on that computer and so on. You know, you need a bit better accuracy than that. So I was basically set during my PhD, I was looking at what technologies could we use indoors to do find your brain location. And if we had that data, how would you process it efficiently to and scalable, easy to make sensor driven systems work. So that was the kind of thing that I did for the next few years. And I built a very big ultrasonic location system. So it was actually something that was you know, how that kind of ultrasonic receiver every two meters all the way across this, this building, but it was an it had tags that you carried around with a bit like beacons today. And you know, they had multi year battery lifetimes. And you could have, you know, hundreds in the in the building and so on. And it was phenomenally accurate, it was really great, you could use the tag as a mouse on a computer screen, you could find exactly where you were to within probably two or three centimeters in 3d. So it was an amazingly accurate system. But it had this downside, which was he was really expensive, because he had so many of these ultrasonic receivers in the ceiling. So I you know, we we deployed that in 1000 meter square building, you know, it was a reasonable sized building and 50 people in it. And, and that's that's basically the, you know, the size of that technology. But it was, it was very interesting to see what you could do with it if you had that kind of data.

    Steve Statler 47:41

    So how many receivers would you need in a building like that roughly?

    Andy Ward 47:45

    Oh, it was something like three or 400? I mean, it was a lot, right, it was a lot. It was a lot of infrastructure. You know, in that kind of research environment, it didn't really matter. It was a, it was more of a proof of principle. You know, if you had this data, what could you do with it? And we built systems around that that did, you know some pretty interesting stuff. So you know, one of the things I thought was really interesting about the active badge work. And the ultrasonic thing that we did afterwards, was that we didn't just do asset finding. I think there's a lot of discussion about beacons and so on. And it always starts with asset finding. And that's a fine application. Right? There's a great thing for that, but but you could do so much more with it, you know, this entire again, it sounds a bit crazy now because the world has moved on in a different way. But one of the most amazing things that they did at the time was because nobody had mobile phones. The receptionist didn't have to forward calls to the system was already hooked up into the PBX. And so when the call came in, it found you know who it was for. It automatically worked out where they were in the building, and you knew that if the phone next to you rang, it was for you. So you didn't have a mobile phone or anything but the phone rang. Pick it up. Yeah. Hey, you know, it was exactly your call. And then you could you could do really interesting things. There are a lot of people at Xerox who had this system as well. And there was a guy there called MC lambing, who was a real pioneer in this stuff. And he built a thing called Forget Me Not and Forget Me Not was an automatic diary. So basically, what it did is it recorded exactly where you were, exactly who was with you, you know, where they went and how the meetings happened, and so on. And the idea was that essentially, you could then go back and look at what happened last Tuesday. And, you know, just as a sort of aid memoir as to what on earth happened last week. Now then we took that and added in integration with your diaries, with you know, other IT systems and so on pulling it together. So essentially, you ended up with a timeline of exactly what happened to you Every day, where did you meet people? What were the photographs that you took when you did that? What documents? Did you look, you know, who was there next to you at the printer last week, you know, when we were chatting over something, you know, it was quite an interesting thing. And we thought at the time that maybe smart offices were the way that this was going to get used. And, you know, in the future, you'd have a whole load of shared stuff around your office. And it would be kind of orchestrated, choreograph by tracking where people were and who is interacting with it, and so on. So that was the kind of idea at the time. It's brilliant.

    Steve Statler 50:37

    And you'd think that now with the smartphone, Forget Me Not like concept could really come into its own because the integration points if Apple or Google got into it, I wonder why not? Maybe it's a privacy thing.

    Andy Ward 50:49

    I think it might be a privacy thing, because, you know, you, you do have to go into it kind of thinking, Well, I'm part of this big hole. And, you know, I'm what I'm trying to benefit from is, is the information shared within that organization and pulled together in a in a, in a timeline that that, you know, is relevant to me. And, you know, I think actually, the the, the kind of mo of the mobile phone is that everything's on here, and it's all yours, and you keep it to yourself. And that's a slightly different kind of mindset from the slightly more expansive. Well, you know, let's put it on the pot and see what comes out. mindset.

    Steve Statler 51:31

    I love the phone. The phone ringing next year, it's like one of those spy movies where you know, the CIA knows exactly where you are. And the President wants to speak to you. And so the phone just rings or promises. Normally, someone behind the president is even more powerful that does that. But

    Andy Ward 51:48

    yeah, I mean, I think, you know, I think actually, although small offices is not where it's ended up. The same concepts are being used with location driven services today in industry, but we'll probably get into that later. So

    Steve Statler 52:01

    yeah, very good. So how did how did that lead to being a co founder of UB sense? How did you do that? And how do you get to be a chief technology officer?

    Andy Ward 52:17

    Well, that I mean, the the, the kind of, say, the demo at this lab was, was really one of, of exploration, and, you know, trying to see whether there were ways in which you could generate interesting technology that had commercial value. So although, you know, the sponsors and the sponsor to the lab over the years were Olivetti and then Oracle, and then at&t, although those were the sponsors of the lab, but they probably weren't going to directly use the technology themselves. But they did fund it, because they got some publicity out of it. And they funded it. Because, you know, the lab had a good history of spinning out companies that took technology that was developed there, and commercializing it. And, you know, obviously, as part of the genesis of these companies, that funding organization will get a slice as well. So this camp, this lab had probably four or five kind of spin outs and a save nowadays, I think it would be viewed as being a sort of incubator accelerator kind of place. But back at the time, it was it was very different from anywhere else. And so, you know, we always had an eye as we were doing this location aware work that, no, maybe there'll be something commercially in it. And as we got more publicity, we were getting probably two or three commercial inquiries a day by the end, from people who had seen what we were doing, and, you know, seeing the publicity about it, and so on, and wanted to use it from, you know, everything from military training through to entertainment, through to, you know, smart offices, agriculture. I mean, there was a real kind of smorgasbord of kind of different things, nothing at that stage where you could say, you know, what, that's a fantastic opportunity. It keeps coming up and up and up. But people were, you know, seeing how they could use that technology and go with it. The best one I ever got, is I got an inquiry from a guy in Transylvania, who had a graveyard that he wanted to survey and it was like, okay, but I'm not sure how to pick up on that. But for real, there was an absolutely genuine inquiry, there's surveying the graveyard. But anyway, so we have this kind of eye on on commercialization, but unfortunately with the the.com bust, we the lab shot before we got a chance to, you know, go out and and start up a company so, together with my other co founders from the lab. We, we found that basically we were, we were on the dole unemployed. And so, you know, when you, when you're unemployed, you have a lot to lose by poking around and seeing whether somebody will fund you to start a company. And so that's exactly what we did. So we took the ideas, I mean, there's some of the general concepts, and, you know, some new ideas we had about technology and so on and look to see whether we could, we could start a company, and that became UB cents in 2002. So,

    Steve Statler 55:33

    so you really have looked across infrareds ultrasonic, ultra wideband, and now and now Bluetooth. So that's, that's quite a,

    Andy Ward 55:50

    I've got a chest at home, like a big kind of chest full of papers about every single location technology you can think of. So, you know, when we did my PhD, I did an awful lot of research into the different ways you can actually do it. And, you know, some of those have been kind of, you know, superseded now by newer techniques, but a lot of it is, is quite old concepts. I mean, the reason I kind of want another reason I got into this is because I picked up a book off my dad's bookshelf once which was called most secret war, it's by a guy called RV Jones. And if you've never read it, it's a fantastic book. But he was in charge of scientific intelligence in the UK, during the Second World War. And, of course, the you know, there was a lot of, to and fro in terms of intelligence gathering and exploitation and so on, in terms of radio, between Germany and Britain, and so both sides, you know, advanced a lot of things, but many of the techniques that we now use for location, were kind of basically first thought of in the Second World War. And so that book is, you know, absolutely is a goldmine of, of how to use that, basically, people came up with them for essentially fighting electronic wars. So

    Steve Statler 57:11

    it is interesting. I mean, how much innovation came out of this war environment. And when we generally think of innovation, people need to be in a place space, it's like all, no, nothing, no wrong answers, and that's the thing. But the flip side is you face, you know, extermination. And you get pretty creative as well. It's another approach to look at, like Israel, where Williams has a lot of its r&d. And that nation is incredibly crazy, but it's surrounded by countries that maybe don't have the best of feelings towards it. Really interesting. Very good. Well, thanks very much, Andy. That's been that's been fascinating. So appreciate the background, appreciate the music and I learned a lot and this was supposed to be the fun bit. I want to thank our hammock for his work on production. Jessie Hazelrigg, our producer, I want to thank you for listening. Please do like us, tell your friends about us. And please join us. For the next time. We meet