2936: Revolutionizing Mobile Security with Metalenz

[00:00:00] Today I'm thrilled to welcome Rob Devlin, co-founder and CEO of MetaLens onto the podcast. Now if you've not heard of them, MetaLens is at the forefront of meta surface optics, a revolutionary technology that is already embedded in tens of millions of devices and

[00:00:21] is scaling up globally with companies like Samsung. And today I want to demystify this technology and also raise a little awareness because you're probably using it without realising it, and explore how MetaLens' PolarID technology

[00:00:35] is promising to deliver superior performance to Android users, but at a fraction of the cost of existing solutions. Well buckle up and hold on tight because I'm going to beam your ears all the way to my

[00:00:49] little home office here in the UK where you can sit down and join me and Rob Devlin in a conversation about all this and much more. So a massive warm welcome to the show. Can you tell everyone listening a little about who you are and what you do?

[00:01:05] My name is Rob Devlin, I'm co-founder and CEO here at MetaLens and at MetaLens we're using our unique technology, meta surface optics, which are a new form of lens and optic to take complicated sensing devices that previously have been locked away in the scientific lab

[00:01:28] or medical labs and shrinking them to a price point and a form factor where we can now put them into millions and hopefully over time billions of consumer devices. As soon as you start talking about scientific labs, and I also read that the technology

[00:01:47] was actually spent outside of Harvard University, always feels like a superhero origin story. So what inspired the creation of MetaLens and how did that technology spin out of Harvard University? There's got to be a big story there, right?

[00:02:01] Yeah, so I think in some ways I missed the boat because if you look at really the most famous Harvard spin outs or the most famous Harvard founders, they dropped out of Harvard when they were an undergrad.

[00:02:13] I made the mistake of going all the way through and finishing my PhD there. But with MetaLens, we spun out of the group of Federico Capasso at Harvard and in his group they had invented these new form of optics called meta surfaces.

[00:02:30] So just briefly what a meta surface is, if you look at traditional lenses like are in webcam, cell phone cameras, anywhere you have a camera or an optical sensing system, these are typically bulky molded shaped lenses where in order to get a good image you might

[00:02:52] need four or five individual lenses. So if you were to go and look in your DSLR camera or in one of the cameras sticking off of the back of your phone for example, there's actually about four or five different lenses in there.

[00:03:07] And so, at Harvard what they were doing was in Federico's group, they were looking at is there a way for me to with just one single completely flat surface, so no curvature like

[00:03:20] traditional lenses, do all of the things that I would need to do with a traditional stack of lenses like we have in cameras today. And so, as they explored this and looked at different ways, they found if they patterned

[00:03:35] a bunch of these nanostructures, so millions and millions of structures that are say a fraction of the width of a human hair, they were able to actually reproduce the functionality of these traditional imaging systems.

[00:03:51] So they were able to take four or five lenses and collapse the function of those four or five lenses into one single flat surface. And then the benefit was also that you were able to make these lenses, so mass produce

[00:04:07] them in the same way that they were making all of the chips that are going into phones. So basically, you were able to leverage this semiconductor manufacturing explosion that has really brought on the information age. So there were two aspects to this.

[00:04:25] It was making complicated optical systems smaller, simpler, but also then having a mass production process that took lenses out of the same way we've been making them for hundreds of years and started putting them in the semiconductor foundry where we can start printing them like computer chips.

[00:04:47] So back in 2016, there had been really nice proof of concepts of this metasurface from the Capasso group and then from other follow-on researchers. But no one had really shown the capability of using one of these new optics, these metasurfaces in order to produce high quality images.

[00:05:10] So when I had joined the group, me and some of my colleagues there at Harvard, we really focused on trying to improve the process and performance and design of these metasurfaces so that you could actually get good images, right? That's ultimately what we want to do with them.

[00:05:32] We want them to be able to be working as well as the traditional optics, just much smaller and much simpler. So in 2016, we finally had through all of this design and iteration, gotten the first

[00:05:47] images with a metasurface where we were actually able to show performance with a single flat lens that was equivalent to a high-end microscope objective that had something like 10 lenses in it. So this was a really big breakthrough. It ended up getting us on the cover of Science magazine.

[00:06:09] So this is one of the biggest journals in academia. And from there, most of us were thinking about going the academic route, continuing on as professors and publishing as much papers as we could and continuing this research.

[00:06:24] But when this breakthrough had come out, we started getting cold calls from some of the largest cell phone manufacturers in the world. We started getting cold calls from investors and from some of the largest semiconductor companies in the world.

[00:06:41] And so it was at that point where we said, okay, well, it's been great pursuing this from an academic perspective, but it seems like now is really the time to go out and see if we can make this into a real product here.

[00:06:56] And what does this look like in taking this technology from the lab and launching it? So that's when we formed Metalens back in 2016 and closed our first round in 2017 and have been really focusing on now putting these into real devices so people can take what

[00:07:15] we did in the lab and interact with them in their phones. It's incredibly cool what you're doing here. And for anyone listening who hears phrases like meta-optics and polarization technology and almost hears back to the futures Doc Brown saying, great Scott.

[00:07:33] How do they differ in simple terms from traditional optics, would you say? Yeah, the biggest difference comes to the size and the shape. So again, if you were to think of the lenses in your cell phone cameras, these are all

[00:07:47] curved and molded and have a three-dimensional shape to them. Especially in some of the highest end phones, they can often have really complicated shapes. So a traditional lens might be something like, let's say, one to two millimeters thick. And then you might have a stack of these.

[00:08:08] So you might have five of those and you'll have spacing between all of those. With a meta-surface, the actual optic that we're making has no curvature to it. So it's completely flat. It's able to bend light and shape light even though it's flat.

[00:08:24] And in terms of the thickness of the actual meta-surface itself, it's hundreds of nanometers. So it's about a thousand times thinner or more than these traditional optics. So these are really, really thin. And that's one of the biggest differences there with the meta-surface compared to traditional optics.

[00:08:49] But another big difference, which you highlighted as well, is that when we think about light, when we think about making an image, most of what we're used to is what our traditional vision system can see. So I'm looking at my computer screen right now.

[00:09:06] My eyes are forming an image of the scene. I see some color. And that's what most of the optical systems that we interact with in our daily lives are doing as well. So, you know, the camera on your phone, whether it's the selfie camera or the rear-facing

[00:09:22] camera, it's really just making an image and giving you some information about the color that is in the scene. But we know that light actually has a whole lot more information in it that traditional optics simply can't access, throw away.

[00:09:42] Or if you are going to try to get access to that additional information in light, this is where you normally need something that is so big it sits on a tabletop in the lab and also so expensive.

[00:09:57] So one of the things that you can do with a meta-surface is actually parse out something called polarization in light. And now, polarization, some people that are photography, let's say, buffs might know about,

[00:10:15] in polarization, you might put a polarizing filter on the front of your DSLR camera and it will let you see more clearly on a hazy day or something like that. So there is some interaction of that that people might be familiar with.

[00:10:31] But really, what polarization in light tells you is it lets you understand what the material you're looking at is made up of. It can tell whether the material I'm looking at is a man-made material or an organic material.

[00:10:48] It can tell you about the shape of the material that you're looking at. So you can actually use the polarization information if you're able to collect it and discern it to tell you the shape of the object that is in front of one of these imaging systems.

[00:11:04] So this is something that meta-surfaces allow you to do in a way from a form factor perspective and a price point perspective that can finally take these really complicated, really expensive polarization cameras and put them into consumer devices for the first time.

[00:11:23] For anyone listening who maybe, I suppose just to bring to life everything that we're talking about here and help them understand how this could work in their world. How is MetaLens technology being utilized in mobile and consumer tech that listeners

[00:11:38] might be aware of that enhances their end user experience? Any examples you can share there? Sure. So one of the first places that we've launched the technology and back in 2022, we made a joint announcement with the company STMicro Electronics.

[00:11:55] One of the first places we launched the technology was in 3D sensing systems that go into consumer devices. So STMicro makes a product called a FlightSense. And what this is, is essentially a camera that allows you to build a 3D map of the scene.

[00:12:18] So the traditional cameras that we have on our cell phones today, most of the ones that we're used to, those are only giving us a 2D image and some color information. But STMicro makes something that will actually allow you to get a 3D map of the room that

[00:12:38] you're looking at or the object that is in front of you. And for STMicro, they are using our optics to, again, replace a complicated stack of optical lenses that go into these 3D sensing systems to improve their performance, to reduce the overall price and cost of them.

[00:12:57] So now they can get into a lot more devices. And in terms of how the end user might then interact with these or how this might impact them, so one of the first ways is getting that 3D map of a room or of the world around

[00:13:12] you, of your environment. So you can use this to build some augmented reality type applications. But another way that these 3D sensing devices get used today, there's sort of two other main use cases. One is actually to enhance the visible camera on the back of the phone.

[00:13:34] So when you take images on your phone and it looks like one of these high-end DSLR cameras with a really nice bokeh effect where, you know, the person is in perfect focus and all

[00:13:46] of the scene around them is blurred out just in the right way, that actually uses one of these 3D sensors. So if you know exactly how far away the person is, you can then make the visible camera do

[00:13:59] some computational photography in order to make these very inexpensive lenses that go into cell phones look more like the high-end DSLR. And then the last place people might be familiar with or maybe not familiar with per se, but

[00:14:16] certainly interact with is for anything that is being used in terms of, let's say, secure facial recognition or secure unlock. You want to have a 3D map of the scene of the object of the person so you know that it's

[00:14:32] not just someone trying to open the phone with a high-quality 2D photo. So this is another place that these type of optics and metal lens is being used today that users are interacting with.

[00:14:46] And I'd love to dig a little bit deeper on that because current facial recognition technology is often criticized for security vulnerabilities. So can you expand on how your solution addresses some of those concerns? Michael Absolutely.

[00:14:59] So there's two pieces to it and this is now for us the second generation of the technology that we're launching. So with STMicro, we launched our first generation of the technology which is sort of on the

[00:15:12] component side where we're replacing a stack of these optical lenses with one of our metasurfaces. And now we're launching a product that we call PolarID. And PolarID is a secure face unlock solution which will allow secure facial recognition

[00:15:31] to proliferate into, let's say, essentially all of the Android phones over time. And now to the security point, on the first point, when we work in terms of launching PolarID with the cell phone OEMs that we're now interacting with, with the cell phone

[00:15:49] manufacturers, we are making sure that that user's data, their polarization data that we use as a way to validate them and keep their identity securely able to unlock their phone, that never leaves the phone. It works in a secure environment.

[00:16:10] Another aspect of this is that if you are going to look at, say, some of the visible facial recognition that has been out there where people are just crawling through the web and looking for all sorts of images of you, we work in polarization, not in the normal

[00:16:34] imaging mode of just, say, the visible cameras. And so that actually means that, you know, if there are cameras outside or the pictures of you that are on the web, those aren't your polarization signature. Those are just intensity images of you.

[00:16:49] So with what we do with PolarID, it actually keeps the user secure in that manner as well. And then the last aspect of this is the Android cell phone manufacturers have wanted to put facial recognition in their phones for a number of years.

[00:17:07] I think many people will be familiar with the fact that Apple has had face unlock in their phone since about 2017 now. But that module they've put in there to be highly secure and have the security standards

[00:17:23] that Apple requires has been too expensive for, let's say, the mass market phones of Android. And so what the Android community has started doing, because their users have been wanting something that they can unlock their phones with just their face, they've started to put out less secure solutions.

[00:17:44] So they've actually compromised on security in order to have the convenience of face unlock in their phones. And what we're able to do at Metalens with PolarID is, again, we're able to take complex

[00:17:59] sensing systems that are normally very expensive and very bulky, shrink them down to a price point and a form factor for consumer. So we're able to now hit a price point that is compatible with these mass market phones. But we don't have to sacrifice on convenience or security.

[00:18:19] It will work indoors, it will work outdoors, it will work in the dark. So that's another way that we end up enhancing security. We now make it that with PolarID someone won't be able to come along and unlock your phone

[00:18:33] because the cell phone that you're using has sort of given a convenient face unlock solution. And with that first generation of your Metasurface technology already in tens of millions of devices, I've got to ask what are the next steps for scaling this technology on a global

[00:18:52] scale there, especially with your partnership with Samsung? I'm not sure if we're allowed to mention that, but again, a huge deal for you guys. Yeah, absolutely. And for us, that really comes to what we're doing now with PolarID.

[00:19:05] And so the partnership with Samsung is on the image sensor. So Samsung is one of the, if not the largest image sensor manufacturer in the world. And when we're making PolarID, ultimately what we want to do is make a product that

[00:19:26] will work in every situation you can think of when you're unlocking your phone. We want to make a product that will work reliably, repeatably, and really protect your security. And a critical part of that is the image sensor.

[00:19:42] So we make meta-optics, which lets us bring this polarization information into the image sensor. Your polarization signature, Neil, so the polarization image we get of you is a unique identifier of you. And it's unique relative to someone coming and even having a perfect 3D mask made of you.

[00:20:06] Your polarization signature will be unique. So in this partnership, what we're doing with Samsung is really we're bringing the meta-optics and now the way that we bring light into this image sensor gives a lot of opportunity

[00:20:20] in terms of joint optimization of the electronics, the sensing with the optics. So that's where the partnership with Samsung really helps us out. And then the way we see now scaling into millions and hopefully over time billions of devices

[00:20:39] is by now enabling secure facial recognition for starting with the, let's say, flagship Android phones. But actually over time, the solution we have is so much cheaper than anything else that is available out there that we even see it displacing the fingerprint sensors that are

[00:21:00] being used in some of the mid-tier or low-end phones. So this allows us to go from the tens of millions of devices that we've been able to launch today to potentially billions of devices because there is another billion phones out

[00:21:16] there outside of Apple that don't have secure face unlock in them today, but would certainly like them from a convenience and security perspective. And I just love how you guys have continuously evolved. And as Metalens moves from a components company to a total solution provider, I've got to

[00:21:37] ask, what are some of the advantages and maybe even some of the challenges that you've encountered along the way? Certainly, it definitely is both of those. And there have been challenges. And I think starting with maybe the challenges, one of the biggest things in going from just

[00:21:53] the component to the complete solution is that we've had to scale very quickly in terms of building software and machine learning teams. So what we're doing now, as I mentioned, and as we talked about in the beginning here,

[00:22:09] Metasurfaces are bringing new information to devices for the first time. So polarization or generally they have unique ways of manipulating the light coming in that you can't do with traditional optics. And this has a pretty big advantage in terms of then co-designing the Metasurface with

[00:22:33] machine learning algorithms, because ultimately what we're doing here is a machine vision, a machine learning type of system. So it's not necessarily the case that for a certain problem, you want to bring the information in the same way that we as humans might want to see it.

[00:22:55] There are ways you can optimize with the algorithms themselves. Now you can optimize the hardware. So we really had to scale up very quickly on the software side. And then the other piece of it is in owning a complete solution, now you are directly

[00:23:14] responsible for the end performance. It's not as if in the component, we might see some challenge with the yield or challenge with the performance of a component we make. We go and we can very systematically track down what has changed in our manufacturing

[00:23:32] process and we can fix it. When you're a solution company, you're impacting the end user and you're dealing directly with the OEMs, which means that you have a lot more responsibility and there are a lot more

[00:23:46] different layers in the stack that you have to peel back to see where the challenges and the problems are arising. So that's been one of the biggest challenges, just that scaling. But then the advantage, which I've already touched on a little bit, is that when you

[00:24:04] are owning the complete solution and with what you can do with Metasurfaces, where we now have a system that we're co-designing and co-optimizing optics with algorithms that are then interpreting the information coming from those optics. It lets you do things that you simply cannot do in traditional systems.

[00:24:27] So I think one of the most exciting things that we're doing now with PolarID besides the beachhead market where we can enable secure face unlock for billions of devices out there, is it lets us bring an entirely new form of sensing to millions and billions of people.

[00:24:50] So today, the vision systems that are on phones today really still mimic our human vision system. So again, we have the visible cameras. Most of us can obviously form an image of the scene with our eyes that we're looking at and get a visible image.

[00:25:10] There are now 3D cameras on some phones, but we also have the ability to perceive depth as humans. But now as Metalens is doing this co-design of the optics and the sensor and the full system, it allows us to bring polarization information to billions of people for the

[00:25:30] first time, which is something we haven't had access to as humans on a mass scale before. So it brings an entirely new form of sensing, almost a new sense to humans for the first time. Wow.

[00:25:45] This pace of technological change at the moment, there's so much happening, I would imagine it's impacting your world too. Are there any predictions for the next revolutionary shift in 3D sensing and machine vision and anything that we should all be paying attention to or something that just

[00:26:01] excites you that you're seeing at the moment? I think a lot of in terms of machine vision sensing, 3D sensing. I think one of the biggest things that I find to be really interesting and there's a lot

[00:26:17] of development and our optics can also help there, our systems that we're making at Metalens can help there is a lot of the, let's say biomarker detection that is being incorporated into wearables. So this is very much a machine vision and sensing type of application where you have

[00:26:37] complicated biological markers you're looking for, whether it be glucose monitoring or blood pressure or things along those lines. And the devices that are often used to make an accurate measurement there are again, things that are big and bulky and live in labs.

[00:27:02] And so this ability now to get the optics smaller, simpler, cheaper, and move them into consumer devices, I think we certainly can help there. And I think that's a really exciting area, just being able to generally understand more

[00:27:19] about ourselves or even with machine vision and 3D sensing in general, being able to understand more about the world around us. So there's so much more autonomy, so many more autonomous devices that are out there.

[00:27:35] And of course the algorithms are important, but if you can bring a richer and better information set into the system for the algorithms to then go and operate on, it makes your algorithms more efficient, faster, make better decisions.

[00:27:53] So I think that's the other place is just rather than simply saying, okay, let's take a really complicated machine learning model and tons of imperfect data. Let's bring better information into the system so that then the machines and our devices

[00:28:10] can make better decisions, faster decisions, and ultimately more complex decisions. RICK Exciting times ahead. And just to offer anyone listening with a valuable takeaway from our discussion today, can you just expand on how MetaLens' solution PolarRTD, how that compares in terms of cost

[00:28:31] and performance to other existing technologies used by companies such as Apple? And also what does this all mean for the average consumer? DENIS Yeah. So certainly I think a good comparison point because of how well they've implemented the solution is what Apple has done with Face ID.

[00:28:50] So in Face ID, you unlock your phone and you barely feel like you're unlocking your phone. It's almost seamless. And that is a very secure system. But in order to achieve that security, it's also a very costly and can be a very large system.

[00:29:09] So if you take something like the traditional secure solution that is out there today, Polar ID can be as much as half the overall cost, less than half the overall footprint. And it can be smaller in the overall thickness of the module.

[00:29:27] And so that's where, again, we see this now being able to bring a secure facial recognition solution to all of the additional phones that are out there on the Android side. So and I think that's a way where now this will be impacting users because fingerprint

[00:29:48] has a number of drawbacks, whether you've just had a slice of pizza and you've got some oil on your fingers, it makes it hard to unlock your phone. It can also end up being if your fingers are too dry, for example, it can be hard to unlock

[00:30:03] your phone. And just that seamless swipe up and have your phone unlock and know that it's going to be in a way that's secure, that it's not just someone with a nice high quality 2D photo or even a high quality 3D mask of you.

[00:30:19] We'll now have this be something that users are directly interacting with and sort of changing the way that they use their phones. But then over time, the other interesting aspect of bringing that polarization information

[00:30:33] to these devices is that, as I mentioned in the beginning, polarization imaging is not necessarily new. It's just that these type of cameras have been too big and too expensive to fit into consumer devices.

[00:30:50] So there are a known set of applications of what you can do with polarization beyond just say secure face unlock, which has some implications in terms of, let's say, health monitoring. So you can use polarization to determine whether skin growths are cancerous or not.

[00:31:09] You can look at things like air quality based off of the polarization signature that you're seeing. So I think that's the other way where over time, after we launch PolarID into this first solution and we continue to build more and more data and more applications, it'll be

[00:31:26] a case where you're able to have your phone be used as something that can shoot a picture off to a doctor and have them help make a diagnosis or take a look at the air quality of the room around you.

[00:31:40] I cannot thank you enough for taking the time to come on here today and demystify the world of meta-optics and polarization technology. But for anyone listening that wants to dig a little bit deeper, find out more information about anything that you're doing here.

[00:31:55] What's the best starting point for all things Metalens? Absolutely. So certainly going to our website metalens.com, there's plenty of information up there in terms of what we're doing. Lots of videos that you can go and we'll have more of an explainer to how the technology

[00:32:12] is working and where else we're going as well as on LinkedIn. Feel free to follow us where we have continuous updates as far as what we're doing next and when it's coming out. Well, thanks so much again.

[00:32:26] I've learned so much from you today around the importance of meta-optics and polarization, how current facial recognition tech and hardware can make consumers vulnerable to potential fraudsters but more importantly, how you're changing that and what you're predicting will

[00:32:41] be the next revolutionary shift and what we should be paying attention to now. Exciting times ahead. I think we're going to have to get you on either later this year or early next year

[00:32:50] to see how things have evolved even further but just thanks for sharing your story today. Absolutely, Neil. Thanks a lot for having me and it was definitely a great conversation. I appreciate it. So as we conclude this discussion with Rob Devlin today, I think it's evident that Metalens'

[00:33:06] meta-surface optics are poised to redefine the landscape of mobile devices and their partnerships with industry giants like Samsung. Incredibly exciting and bringing this cutting-edge technology to a global audience. But I'd love to hear your thoughts on anything we talked about today.

[00:33:21] Please email me techblogwriteratwork.com, Twitter, LinkedIn, Instagram just at Neil C. Cubes. Let me know your thoughts. Other than that, I'll be back bright and early in your podcast feed tomorrow morning. So thanks for listening as always and until next time, don't be a stranger.