LiDAR Measures the Truth of the World

Executive Summary

Angus Pacala, Co-Founder and CEO, Ouster joined Grayson Brulte on The Road to Autonomy podcast to discuss the launch of their REV8 native color LiDAR and its significant impact on the autonomy economy. The conversation covers the intense engineering rigor behind functional safety, the strategic acquisition of StereoLabs, and how unified sensing is poised to accelerate physical AI, humanoids, and defense applications.

Key The Road to Autonomy Episode Questions Answered

The Road to Autonomy Topics & Timestamps

[0:00] AUTNMY AI

The episode opens with the premise that billions, and soon trillions, in value will be created within the autonomy economy. The key to capturing this value is uncovering signals before they become mainstream headlines and move the markets.

[0:36] Changing LiDAR Industry

Angus Pacala discusses the continuous consolidation of the LiDAR space since 2015, noting that marketing alone cannot sustain a hardware company. Ouster separated itself from the pack by relying on high-quality, iterative digital products rather than marketing tricks for safety-critical applications.

[03:56] Introducing REV8

Ouster unveils the REV8, the realization of a 10-year vision to create a native color LiDAR sensor. Angus describes this as a “holy grail” for robotics, offering perfectly synchronized color and depth from a single device that is fully engineered to meet functional safety standards.

[13:42] Building Trust with Safety-Critical LiDAR

Angus Pacala details the intense engineering rigor required to achieve functional safety certifications, such as ISO 21434 and ASIL B. The process involves scrutinized code, redundant systems, and self-diagnostics between measurements to ensure absolute trust and reliability in safety-critical applications.

[17:53] Why Custom Silicon is Ouster’s Moat

Ouster’s custom in-house L4 and L4 Max silicon chips serve as a massive competitive advantage, enabling exponential year-over-year performance improvements. This silicon architecture is the foundational tool that allowed Ouster to integrate native color and extreme digital signal processing directly onto the chip.

[25:33] Color Science Behind REV8

To perfect the REV8’s output, Ouster partnered with Fujifilm and DxOMark to combine raw spectral accuracy with subjective human color perception. The resulting sensor handles massive dynamic range (116 dB) and features cinema-quality 48-bit color depth, matching or beating modern automotive cameras.

[33:28] Can Color LiDAR Replace Cameras?

Angus Pacala confirms that native color LiDAR allows many customers to simplify their systems and completely replace traditional cameras. Colorized point clouds add critical context to raw measurements, enabling robots to distinguish crops from dirt in agriculture or detect the state of a traffic light.

[36:36] StereoLabs Acquisition

Ouster acquired StereoLabs to quickly achieve its vision of unified sensing and perception. By combining StereoLabs’ stereo cameras with Ouster’s LiDARs, developers can now seamlessly fuse data into unified, colorized point clouds using free software development kits (SDKs).

[40:07] Ouster as a Sensing Company

Ouster is positioning itself to be the foundational sensing and perception layer for all physical AI, playing a role similar to NVIDIA’s in computing. With their comprehensive suite of sensors, Ouster is perfectly positioned to supply multimodal systems, from manufacturing robots and autonomous vehicles to humanoid robots.

[49:46] Defense Applications

While military automation is still in early development, Ouster is finding immediate success in drone-based surveying as the first Blue UAS-certified LiDAR payload. Angus is also highly bullish on the future of rugged, robotic leader-follower “pack mules” designed to carry gear and support troops in the field.

[52:14] Future of Ouster

The company’s ultimate goal is to lead the world as the premier platform for unified sensing and perception. Angus wants Ouster to be a brand synonymous with trust and safety, ensuring people feel completely secure when interacting with any autonomous vehicle or robot powered by their technology.

Full Episode Transcript

AUTNMY AI

AUTNMY AI: Billions and soon trillions in value will be created in the autonomy economy. By the time a trend becomes consensus, the alpha is already gone. The gap between uncovering signals and reading headlines is widening fast. When it’s a headline, it’s no longer a signal. Enter AUTNMY AI, we decode signals before they move markets, giving you the edge while everyone else tries to decipher the autonomy economy. AUTNMY AI, your models, our intelligence. Visit AUTNMY.AI.

Changing LiDAR Industry

Grayson Brulte: Angus, I have to let you in on a little secret here. You are the only guest where I get at least an email or a DM every week asking when you are going to come back on. So thanks for coming back on. You are highly in demand. You were last on in February 2025. We are not going to let that much time pass, but what has changed in the LiDAR industry since you came last on in February 2025 to today?

Angus Pacala: Man, well, first, thanks for having me back. This is always a fun conversation. A lot has happened internally at Ouster, and especially in the last couple of weeks, actually since the beginning of the year at Ouster with acquisitions and big product releases. But the LiDAR industry generally is a continual thinning out of the industry. I am always amazed how long it takes for a new technology industry to shake out. I think back to 2015 when we founded Ouster, there were over 100 LiDAR startups. Basically from 2015 through 2020, there were over 100 startups in this space, and the thinning out takes decades. We saw a couple of companies really fall by the wayside, even declare bankruptcy through last year, and their assets being acquired. I always wonder to the average person trying to make sense of this space, if it is possible for the average person to make sense of this space. I have kept it in my head, kind of watching the ins and outs of the industry over the last decade. But it is a convoluted set of companies that have very good marketing. Ultimately, Ouster has definitely separated itself from the pack in the last five years, probably since going public. But there is still a bunch of competitors making noise every single earnings cycle. So it is a fun space.

Grayson Brulte: Marketing will only take you so far. It comes down to the quality of the product. Has that been one of the defining aspects that allowed Ouster to grow and prosper as a public company?

Angus Pacala: Exactly. This is something I say internally, you will never trick someone into buying a multi-thousand dollar LiDAR sensor or a LiDAR sensor of any kind really that is in a safety critical application. You will never market them into buying a product. LiDAR is too critical to these robotic systems. It is too much a part of the safety scope. There is too much liability, and the problems we are trying to solve are far too hard to ever trick anyone into buying your product. We understood that very early on. We got immense feedback that was negative, critical in our early products that we released all the way back in 2018. It was painful to hear at the time, but instead of just giving up at that point or refusing to listen to customer feedback, we iterated and we improved our products. That was kind of the whole idea of the company after all, it was digital products. You can iterate them, you can improve them over time. And here we are, Rev 8, our eighth iteration of our products, and these are the best LiDAR sensors you can buy in the world, bar none.

Introducing REV8

Grayson Brulte: And their color. I want to point that out, their color. How did you cross that line to make them color?

Angus Pacala: Let’s talk Rev 8. We haven’t released a product at Ouster on the hardware side for three straight years. The last time we released products was the Rev 7 product line, which we released in early 2023. Internally we have not just been sitting around. Ouster works really hard, and we have been exclusively focused on this one product release, Rev 8 with native color. In many ways it is not just a three-year journey, it is actually a 10-year journey for me and my co-founder, Mark. We actually in our original pitch decks outlined the goal of building native color LiDAR. So this goes back a decade. We had this vision that the holy grail robotic sensor would have color and depth all coming out of one device, perfectly matched, synchronized, and seamless to the user. It just makes sense, if you think about what a robot needs to describe the world around it, it is obviously a combination of color and depth. We started with LiDAR because that is really the critical enabling technology for so many safety critical robotics, but we always wanted to add color. We actually had multiple generations of our silicon, including Rev 7, where the silicon itself had color sensing in the silicon, but we just couldn’t quite get the right combination of performance and capability into Rev 7 for it to be worth shipping. So we waited, and we made it better, and we made this native color sensing better and better. Now Rev 8 is the first time that any LiDAR in the world has this native color technology. We are going to get into it. There are a lot of details here, but this was not an afterthought. This was a 10-year vision for me and my co-founder, Mark, and it feels really good to finally have it out and see the incredible reaction for roboticists that just get it. They just see that this is the obvious path forward.

Grayson Brulte: 10 years ago, what did you see that made you and Mark say, “Okay, color is the path forward”?

Angus Pacala: We pride ourselves on really understanding our customer applications. I spend a lot of my time programming robotics and talking to customers. I thought about getting a computer vision degree after my bachelor’s degree. So I have a core interest in computer perception. I had a lot of fun in the early 2010s following the Kinect. There was this community of people and roboticists built around the Kinect sensor. The Kinect was this toy basically for the Microsoft Xbox that was one of the first widely available decent quality RGBD or color and depth consumer depth cameras. It was so cool. It just seemed obvious that if you could pack that kind of colorized point cloud capability into a device with the raw performance and all-weather outdoor long-range sensing of a LiDAR, that you would have something obviously better and more useful for the customers and the robotics community. I would say actually that the Kinect journey that I watched happening online was a factor. There were other things. Willow Garage was using a lot of RGBD back in the day. There was Project Tango at Google. These were groups using really short-range RGBD depth sensors that were colorized. But they weren’t LiDARs. They weren’t the kind of high-performance devices you need to make modern physical AI safety critical robotics. We brought that kind of context to the plate when we set out this vision 10 years ago for Ouster to build native color LiDAR.

Grayson Brulte: Kinect to me was one of those moments. I played with Kinect and the Microsoft team built that really great jet ski game where you move your hands and you move the jet skis. But I didn’t really play the roboticist then. I was advising a gentleman in media. He had a startup and his second or third one after a few exits. When you can move your hand this way to control the screen from a user experience standpoint to get the news that you wanted, it just seems like that Kinect had a lot underneath that technology that inspired you to go down the robotics route and inspired me to go down this media hole. Do you look back and say that’s one of those critical moments in engineering history that led to a variety of sub-industries?

Angus Pacala: Yes and no. There was so much promise for the Kinect and that first wave of depth sensors. Leap Motion was another one, not colorized sensing, but 3D sensing and basically human computer interfaces that were more natural. The reality really didn’t quite live up to the promise from 2010 through 2015. The Kinect was canceled and turned into more of an OEM device. Leap Motion failed as a company ultimately and just didn’t have the right product market fit. It was a first wave that showed what might be possible in robotics, but things just weren’t quite aligned, the software wasn’t good enough, and the hardware definitely wasn’t good enough for all the applications that Ouster serves. The challenge is it is not just about making native color point clouds. It is about making them in a device as performant as a LiDAR, as rugged as a LiDAR, and without caveat as an Ouster LiDAR today, and also now functionally safe. That’s the other headline in our Rev-8 announcement that didn’t get a lot of airtime in our release video, all of our Rev-8 products are not just native color devices, they are functionally safe devices. They are engineered from the ground up to be a device that you can risk your life on. When you ask what took three years to get a product out at Ouster, it’s a combination of native color and functional safety. Guaranteeing to our customers that this is a product that will not fail, or if it does fail, we’ll let you know within 100 milliseconds so that the robot can emergency stop and keep you all safe. We’ve taken that really seriously and putting the combination of those two things into a device is crucial. Basically you need all of these things to align to make practical physical AI robotics scale and deliver on that promise that the Kinect potentially started in 2010.

Building Trust with Safety-Critical LiDAR

Grayson Brulte: It started ideas and hits. The technology didn’t mature, but it obviously seeded a lot of ideas. Your ISO 21434 standard compliance, you have that as well. That is a big deal.

Angus Pacala: It is a huge deal. The entire ASIL B, SIL 2, PLd effort required, I don’t think that the average person really understands. If you haven’t been an engineer in the thick of developing functionally safe devices, you really don’t have appreciation for the level of scrutiny, detail, and effort that goes into this. The engineering in the Ouster Rev 8 LiDARs is much more akin to the engineering in a rocket that takes astronauts into space than it is into a consumer device like a Kinect or a security camera. The orders of magnitude difference in effort is probably 1,000 times harder to get a product out that is functionally safe. Because the core idea is we have to be able to sell every last sensor that comes off our line and it needs to be guaranteed to a standard of quality and reliability and self-diagnosis and performance that is way beyond a normal warranty. If we had a 1% warranty rate, that’s a very good warranty rate for a product because something just isn’t quite right about the product. That’s nowhere near good enough for a functionally safe device. You can’t have 1% of your products out of spec and telling a customer that’s using these in a safety critical application the wrong thing. We completely shifted the perspective of the entire engineering team as we designed these Rev 8 products to a world where we were all thinking, “We need to be able to trust our own lives on the product that we are designing. We need to, before we prove this to anyone else, we need to prove this to ourselves, then prove this to TUV SUD, our safety assessor.” Obviously not just to prove so that we can get the certifications from a third-party assessor, but then we can go out and sell these products in a certified way to our customers. Every last detail, every last component, and every last line of code has been scrutinized, tested, verified, and then ensured continues to fall into spec as we produce these products at scale.

Grayson Brulte: That is the engineering technical side of the equation. But the other equation that in some cases matters more is the brand. It is the Ouster brand. When you stamp that brand on that LiDAR and you ship it out the door, if there is an issue or something fails, that is not good. You can have the world’s greatest engineering, but it comes down to the brand. It seems to me what you described is that you want to build the world’s best LiDAR and you stand behind your brand. That is what it says to me.

Angus Pacala: Exactly. Once you cross this threshold into safety critical LiDAR, the reputation of the company is as important as anything. You really have to prove this to your customer base, though. This is Ouster’s first foray into functionally safe devices. One of the reasons you have a third-party assessor is because you have no reputation for shipping these types of devices. There is no way to avoid the hard work of building this reputation over time. It is again why we took three years to develop these products. Having a misstep out of the gate and not having a high-quality device and any question in a customer’s minds of whether this adheres to the absolute rigor of the functional safety standards means that we won’t be successful in these markets. The whole point of a customer buying a functionally safe device is so that they can put complete trust into these LiDAR sensors. In the same way that you and I buy a car and we put complete trust into the brakes. The brakes are a single redundant system that just cannot fail and is a functionally safe system in your cars. If your brakes ever fail, which they don’t, it really calls into question your trust for that system. Seeping doubt cannot be something that we sow with our customer base. Again, every last detail has been scrutinized in these devices down to measuring every voltage rail. Every transistor on the silicon, for instance, has to be doubly redundant checked. One of the core concepts actually of functionally safe devices is being redundant and self-checking. Through redundancy of basically having two things do the same thing and comparing the result internally so you can do fault checking or self-testing the components in between real measurements or real work that you are doing as a LiDAR, you run little self-diagnostics. That goes all the way down to the custom L4 and L4 Max silicon chips that we have developed. The actual transistors on that chip need to be tested and self-checked between every LiDAR measurement. In many cases, we have redundant transistors on the chip so that there are two different systems doing the exact same work on the silicon so that we can check for faults. Scenarios that we ask ourselves about are what happens if a neutrino or a gamma ray from space comes and flips a bit in the ASIC? Well, what do you do? You can’t have this LiDAR output unsafe data, so you need to have redundant transistors so that that gamma ray only hits the one transistor, flips the one, and then the other redundant transistor gives you the fault checking and you have a diagnostic that you can send to the customer. This gets really detailed. I could nerd out all day on it, and I am happy to go deeper, but it is really incredible what companies that build functionally safe devices have done to meet that bar.

Grayson Brulte: It is a very high bar to meet. I’ll give you credit from a strategy standpoint, from the TUV SUD standpoint, that potentially opens the EU market up because a lot of the EU regulations are benchmarked to TUV SUD.

Why Custom Silicon is Ouster’s Moat

Grayson Brulte: You mentioned the silicon. You have your own custom silicon. Is one of the reasons that you decided to build your own custom silicon was to meet the Ouster safety bar?

Angus Pacala: The silicon development at Ouster is so core to everything we do. The answer is yes, absolutely. One of the reasons we do custom silicon is because then we can be at the forefront of this functional safe technology. But it is the core founding strategy of Ouster to build the first digital LiDARs. “Silicon eats industries” is what we said in our original pitch deck. Any industry where silicon becomes the major technology that drives the underlying product performance and economics is absolutely dominated by silicon. Look at cameras, that’s the best example. An optical sensor used to be film and analog with thousands of components, and digital CMOS cameras completely exploded the market. They didn’t just capture the professional and prosumer market, they dominated and expanded what cameras are even used for. That was all a CMOS silicon technology that made that happen. Silicon eats industries, GPS receivers, even silicon oscillators is something that’s happened in the last five years. That was Ouster’s core thesis, and it allows us to iterate and improve the performance. It allowed us to put native color on the silicon. We never would have gotten to native color without doing our own silicon designs. And it gave us the path to functionally safe devices. There are few free lunches in the world of hardware development, but in a sense, silicon gives you a free lunch to pack more capability, performance, features, and safety onto the same silicon die area each and every year because of the amazing march of improvement in the underlying transistor technologies that we have access to.

Grayson Brulte: To me, you control your own destiny when you control the silicon, because you can enhance performance, you can build the color in it, you can put the safety in it. When did Ouster first develop and commercialize your own silicon?

Angus Pacala: From the very start, we knew we were going to do our own custom silicon. This goes back to how hard it is for the average person to read about the LiDAR industry and make sense of who’s telling the truth and who’s embellishing. At this point, every LiDAR company says that they do some sort of semiconductor development. Ouster looks like just one of the mix. But the difference is Ouster founded the company on silicon design. Our very first product we released in 2018 was soup to nuts an in-house silicon design. We shipped a fully custom piece of silicon that was the L1 chip back in 2018. Now we are on our fourth generation architecture. But we learned the hard way to get to even that first 2018 product. We did three different tape-outs that were test chips from 2015 to 2018 to get the underlying silicon architecture right. Those were a very difficult learning process. There’s no way to learn other than by fire and make some mistakes. We’ve actually done eight unique architectural tape-outs at Ouster over our 10-year history, and we’ve gotten really good at silicon design. We have an in-house silicon design team and four major architectures that we’ve shipped to our customers, L1, L2, L3, L4, and L4 Max. This is the first generation where we’ve introduced two chips into the architecture, the L4 and the L4 Max. We are doing silicon design and shipping at scale. Our silicon is much more complex than a lot of the other competitors in the space. It’s not just a little sliver of silicon doing photosensing. We have 40 giga MACs of digital signal processing on the chip, huge banks of memory and SRAM, and 20 gigabit per second data off the chip. These are incredible devices with a big silicon area that is the core driving force of all the performance and safety and unit economics of our LiDARs.

Grayson Brulte: If you compare Ouster to Apple, this is my opinion as a consumer of Apple, one of Apple’s great advantages are their M chips. They are phenomenal, fast, reliable, and if you go look at public data, you can’t buy a Mac Mini today because of the demand for the chips to run agents, and that is a huge competitive advantage to Apple. Do you feel, in your opinion from Ouster, that your custom LiDAR is your M chip, your huge advantage in the LiDAR market?

Angus Pacala: Absolutely. The dynamics are in every market, the reality is you have competitors, and what tools are you going to put into your toolbox to keep ahead of all your competitors? There is no better tool than an in-house silicon design team that’s competent and can deliver year in and year out. Because there’s no other hardware technology like silicon that has an exponential improvement curve. Everything else in hardware is staunchly in a linear improvement curve. Look at battery technology. Batteries improve 5% per year. It’s an amazing technology, but it’s 5% per year. Optics, same thing, 5% a year improvement in coatings or numeric aperture. But silicon doubles in performance every year or two years, even today, especially with the nodes that we are operating with. So it’s a tool in our toolbox that’s a huge competitive advantage versus the rest of the industry. You have to almost put blind faith in the fact that as a company, you are going to continue to invest in new silicon each and every year. This just never stops. And if you step back and look at what that’s done for a company like Ouster over a course of 10 years of chip tape outs, eight chips, four architectures. It is night and day difference between the products that we shipped in 2018 and the products we have today. And if you look at that at a company like NVIDIA over the course of 30 years, Jensen Huang has said that he put blind faith in just continuing to tape out more and more impressive chips, packing more capability into their GPUs. To the point that in the late ’90s, some of their customers were telling them they don’t need more capability in their GPUs, and he ignored them and said, “I’m sure once you get your hands on it, you will.” Compounded over 30 years, you get a company like NVIDIA.

Color Science Behind REV8

Grayson Brulte: And you go back, I remember this. I was a video gamer and they were competing against RTX. This is when Doom and Duke Nukem and all those first shooter games were there, and we always wanted the new graphics and we were building them. But the thing about NVIDIA is they never stopped building, and Jensen never said, “Okay, I’m going to rest on my laurels and we’re going to have this.” They kept building and building and building, and look where they are today, one of the world’s most valuable companies. And it’s a clear denominator, no matter what side of the aisle you’re on, is that silicon unlocks greatness. It is powering the global economy. And since the new LiDARs are color, what type of perception are you getting? What type of depth? I read in the press release that you had a relationship with Fujifilm, and I got all excited about that because they make really great color. What type of color and perception are we getting?

Angus Pacala: This has been a really interesting learning process for Ouster internally, as shipping native color was not an afterthought. We had to go partner with amazing companies like Fujifilm and gain access to their color science that they’ve developed over decades, starting in film and moving into digital photography. We partnered with DxOMark, a world leader in camera testing and calibration, so that we brought rigor into the testing of our native color technology. This has been a journey to build this core competency at Ouster. I have gotten the question again and again from customers, “Well, how do you know how to build a camera?” This is going back years as we were confidentially sharing some of the details on the Rev-8 roadmap with customers. The answer was that we learned through hard work and through partnering with the world’s best companies. It’s interesting because LiDAR measures the truth of the world. When you shoot a ray out at an object and time how long it takes back to measure the distance, there’s only one true answer to the distance to all objects in the world. Ouster’s goal is to go measure that truth. So there’s a level of rigor that we can apply to testing and validating our products that’s very straightforward in a sense. It takes a ton of engineering talent to actually go do that at scale. But there’s truth, and we’re measuring it. When it comes to color, color is an interesting combination of true truth in the spectrum. There’s a true spectral response from everything in the world that dictates its color. But actually, there’s a layer of human perception, which is more subjective, that dictates what a good camera is from a bad camera. Human eyes, the rods and cones in your eyes, the optics of your pupil, in combination with your visual cortex, actually dictates what makes a good camera. Good cameras are much more cameras that replicate the human perceptual system rather than cameras that just purely measure spectrum. We decided we are going to give customers the best of both worlds, so our native color technology is one of the best pure spectral cameras measuring the truth of the spectrum and the environment that you can buy. But we also spent the time working with Fujifilm and DxOMark to add a layer, an optional layer that we’re giving our customers, that massages color to match the human perceptual system. That’s been a journey, but it requires a huge amount of detail and understanding, and this is where the term color science really comes into play and Fujifilm’s color science. Every camera company markets their color science, which is the process of taking raw measurements and making them adhere to the human perceptual output, so that when a normal person looks at a photograph, they say, “Yes, that replicates what I actually see in the real world,” versus it being something that looks shifted in color or something like that.

Grayson Brulte: My first job when I was a kid, I worked in a camera store. We were a Kodak store, but don’t tell anybody, we still used Fuji machines. I wasn’t the owner, but we got to learn with it and play with the different chemicals and films. Fuji had the Velvia, which would shoot at ISO 50, and then the Fuji 100. It was real interesting when you would shoot the Fuji Velvia, the slides at 50, the amount of vividness that you would get when you would shoot the 100 or if you would go shoot a Kodak Portra and a Velveeta film, and they shot at 160 and 400. You would see different variations in the color of how it would perceive the world. Different applications were used for different reasons. Was the goal when you were working with Fuji to say, “Okay, the vivid stuff is great if you want to have a photograph and put it on your wall, but we need actual reality of what’s being seen.” Was that one of the things when you worked with the color science team that you were trying to achieve?

Angus Pacala: Absolutely. So, Fujifilm has done a great job of building color filter technology that doesn’t embellish the data. It gives you accuracy. And then there are ways to take that accurate data and further massage it into a format that is useful for ingesting by a machine. So that’s more Ouster’s job is to take great raw measurements that are actually more dictated by the Fujifilm color science and the color filter technology that they’ve developed, and then with digital signal processing or image signal processing, further massage that data with computational imaging into the final output for customers. Things that you have to deal with in that process are high dynamic range. The human eyes and most cameras today cannot deal with the deepest shadows in the same scene as bright lights from a stoplight, or let alone the sun shining right into a camera. How do you make sense of that dynamic range and first of all, have the hardware that can measure dynamic range to that level, and second of all, then represent it to a customer in a way that’s useful for a machine to interpret. We spent a huge amount of time on things like this. There’s no other native LiDAR out there. But when you compare our native color LiDAR output to a normal camera, we are at the top end in terms of color accuracy, bit depth, 48-bit color depth is like cinema camera quality bit depth. Dynamic range, 116 dB of dynamic range, and then one lux all the way to 20 million lux sensing capability. That’s a huge range of luminosity that we are able to measure with our LiDAR. Native color in Ouster LiDARs is not just a tack-on or bolt-on feature that’s a nice-to-have. It is, pound for pound, an exceptional camera in its own right, and you compare it to a modern automotive camera and it’s as good or better in all the ways that matter for a camera in that domain. A lot has gone into native color for sure.

Can Color LiDAR Replace Cameras?

Grayson Brulte: Which unlocks a lot of value, a lot of applications in theory from automotive or machine physical AI applications. Could you see your color LiDARs potentially replacing camera?

Angus Pacala: Certainly there are many cases where customers don’t need a camera anymore. If you’ve seen the release imagery from our marketing release last week, most people that have seen the video and that aren’t steeped in the LiDAR space and are non-technical have just asked me, “Okay, I see that that’s the camera,” and they basically confuse the LiDAR output as a camera output, and I’ve had to explain time and again to all my relatives, “No, no, no. What you’re seeing is the LiDAR generating all this RGB imagery. It’s not a camera that we used as marketing material separately.” So yes, in a lot of cases we’re just going to have customers that can simplify their systems around native color LiDAR. And if you ask where the LiDAR industry is going, I think universal adoption of native color LiDAR is the route forward. It makes no sense to be operating in this black and white world that we have been for the entire history of LiDAR. Once you see colorized point clouds and you get your hands on them, and we’ve actually released a large number of real datasets that you can just go download. We’re putting the proof in the data for our customers to go test out day one. But once you get your hands on that, there’s just no going back. Every application has some level of context and information in the color data itself. In agriculture, you really want to know where your plants are versus everything else. The color green is a very important input to machine learning algorithms versus dirt. Brown versus green versus the sky versus most objects in the field are going to be a different color than the environment that they’re in. The canonical examples that we provided in our marketing materials, stoplights. Every LiDAR that’s ever been made could not sense the state of a stoplight. It can detect that the stoplight is there and locate it in 3D space perfectly, but you don’t know the state of the stoplight. So you can’t build historically a robot with just LiDAR sensors that can navigate a surface street, and now you can. Native color gives you the chance and the opportunity to build a unified sensor around just native color LiDAR for the very first time with no other LiDAR or no other sensor needed. Now does that mean that every customer is going to rip out all the cameras in the system and just use native color LiDAR? No, because there are great uses for cameras still for really high resolution sensing, reading text on a sign that maybe the Rev-8 sensor still can’t do. But there are plenty of applications that are going full just Rev-8 native color LiDAR for sure.

StereoLabs Acquisition

Grayson Brulte: The agriculture is interesting. We have covered agriculture here a lot in that you are starting to see automation and autonomy in agriculture accelerate. Now you are putting color LiDAR into the mix. Do you see color LiDAR acting as an accelerant for the adoption of physical AI?

Angus Pacala: I really do. I think that this is a game-changing kind of paradigm-shifting moment for safe, flexible, capable physical AI. Because again, we’re combining two paradigm-shifting capabilities into these systems. It’s functionally safe and it’s a native color LiDAR. You really need the combination of those two to make physical AI a reality in the real world. At Ouster I’ve always prided myself on building the real production-grade ruggedized hardware that customers need to get their products out in the field. Ouster solves the hard problems so our customers don’t have to, at least when it comes to sensing and perception. We’ve done that in droves with Rev-8. We’re taking a huge amount of the effort and the difficulty of fielding physical AI systems by making rugged, functionally safe colorized systems. So definitely I think this is going to be the technology that powers this next wave of actually deployed physical AI.

Grayson Brulte: It is going to be fascinating to watch what your customers use the Rev-8 for to deploy, and it is not just color LiDAR that is keeping you busy. In February, you acquired Stereo Labs. What is that going to unlock for the company?

Angus Pacala: Yeah. So this is the beauty actually of the Stereo Labs acquisition, is Stereo Labs is an incredible team that’s been building world-class stereo cameras for over a decade, based in Paris, incredibly passionate. And guess what the output of stereo cameras are? It’s colorized point clouds. It’s native color point clouds. We announced to much fanfare that when we acquired Stereo Labs, we would build unified sensing and perception. Well, what’s the most important thing to unify at the very start? It’s the data that’s coming out of our devices that we sell to customers, and that means colorized point clouds, native color point clouds. So we rapidly closed a gap in about three months from acquiring Stereo Labs to announcing Rev 8. We now sell unified sensing. Every product that we sell can give you colorized point clouds. And that gives a path for software engineers to write unified algorithms that can take advantage of the presence of both of those sensing modalities at all times, right? The only way you get convenience and unlock new capability is if you have guarantees on the type of data you have access to at all times. So the vision that I have for the company or the combined companies is that customers can pick and choose off this smorgasbord of sensor products that we sell, stereo cameras, mono cameras with neural depth and Rev 8 colorized LiDARs, and fuse them all in software seamlessly into one colorized point cloud that is a unified way of dealing with this data and allows for a really high performance path forward for physical AI software. And this is pretty obvious to a certain set of customers that have immediately reached out and been like, “Oh, I get it now.” Colorized point clouds is the holy grail of a robotic engineer’s sensory input. And now Ouster offers that across our entire product lineup, whether it’s the Stereo Labs products or the Rev 8s.

Ouster as a Sensing Company

Grayson Brulte: Is Ouster then becoming the sensing company for the physical AI economy?

Angus Pacala: That is the vision, yeah. I mean, someone needs to do it. Some company needs to provide the foundational sensing and perception of safe physical AI. This isn’t just about getting prototypes out. It is about building robust, safe systems at scale. And it’s what NVIDIA has done for the compute side. Someone had to do it. This doesn’t just happen through random diffusion of ideas through the economy. Companies have to come out with a vision and build things that really work in the real world. So our vision is to be the sensing and perception foundational layer for all of this robotics and physical AI work. And I think we are well on that path, right? We are a leading provider of both the camera and the LiDAR technology already. And the recent product announcements actually both from Stereo Labs and Ouster are just catapulting us even further ahead of our competitors.

Grayson Brulte: Yeah, because the best part is you’re making these strategic moves and you’re doing that as physical AI accelerates. You can’t turn on CNBC or Bloomberg or open the Financial Times or still read print without seeing a mention of physical AI. And a lot of the credit goes to Jensen and NVIDIA, so thank you for bringing robotics from a lab to the cool world. So thank you for that. One of the interesting things about Stereo Labs was they have a very passionate and dedicated developer community. Will Ouster continue to support that community?

Angus Pacala: Of course. That’s actually one of the things that I noticed and made me confident that Stereo Labs was truly in the lead, was the passionate customer base that they have. They have over 10,000 customers. You don’t trick anyone into buying these products. 10,000 customers doesn’t happen overnight. They’ve actually built that over time. They have great intuition about what this robotics customer base needs. And they have things like the developer community that’s incredibly active and open, by the way. Active and open. You can go look and read all of the good, the bad, and the ugly. That’s the kind of thing that I look for, is a company that’s not hiding behind logins or paywalls or something like that. We are building a platform, and part of a platform is focusing on the developer experience and the developer community. So now, even before we acquired Stereo Labs, Ouster started a developer community in the same exact style as the Stereo Labs one. We want to make sure that we have open and direct communication with the developers that are working with our hardware and increasingly working with our software development kits. Stereo Labs with the ZED SDK and Ouster with the Ouster SDK. These are free SDKs that we want to support thousands and thousands of customers with, and I think that’s resonating just when you look at the activity on these forums.

Grayson Brulte: Again, and as you said, they are public. Anybody can go through the forums. I have gone through them, and I was like, “Wow, okay. This is starting to make a lot of sense here.” And developer communities are a great accelerant for a company, and you get a really good pulse of where that company’s going from a product perspective. Because if the developers don’t want to build your product on top of it, you really don’t have a business, because developers are what is driving this. And one of the big developments now that you and I are aware of, and the majority of this audience is the rise of humanoids. Humanoids are coming fast and furious. With the acquisition of Stereo Labs, does that act as a potential accelerant for Ouster’s move to service that market?

Angus Pacala: Definitely. Stereo Labs already has carved out a presence in the humanoid market. They haven’t announced that many customers, but if you look around at marketing announcements, and there’s a lot of humanoid companies, sometimes you will be able to infer which products are being used in the stereo systems that most humanoid robotics uses. So yeah, Stereo Labs has done a great job. The founders there have great intuition, great product sense, great technical acumen, and they have been at the forefront of this wave of physical AI robotics and humanoids in robotic gripping and robotic arms that are now much more flexible than ever before and really leverage the same type of unified models that the humanoids do. Definitely the types of LiDARs that Ouster sells today, the Rev 8s, are right at the edge of what you would put on a humanoid robot. And I think it is yet to be seen whether humanoid robotics adopt the type of incredibly high-performance LiDAR that we sell, just given that they are at this threshold of the kinetic energy of the system can hurt someone. There is safety involved, but it is much lower than a car or a big industrial machine like a forklift. And the sensing needs are much shorter range. Although I say that, and then we just had a great example from the Beijing Marathon and they had a robotic half marathon, and actually first, second, and third place in the robotic half marathon used Ouster LiDARs. And it just goes to show you that once you get a robot like a humanoid moving at 20 miles an hour or 15 miles an hour, whatever they were running, all of a sudden your sensing needs go way up. Both because the kinetic energy in the system is much higher and it becomes a safety thing, and also when you move faster, you have to sense much further out. And this was an outdoor environment, so all three first, second and third, the podium was swept by humanoids operating with Ouster Rev-7 LiDARs. The point being that humanoids have not universally adopted LiDAR, but they have universally adopted stereo cameras and monocular cameras, and Stereo Labs sells world-class technology and products on that front. And they actually just announced another new product, the ZX Nano, in the last quarter, which is a purpose-built low stereo baseline wrist-mounted camera that gives great view of the actual hands and robotic grippers that are used on all these platforms. Because it turns out more sensors, way better. You ingest all this data by a big neural network and you get better results, and that includes cameras that are dedicated to very short range but highly detailed and accurate stereo vision of the grippers themselves.

Grayson Brulte: Well, you are clearly positioning the company for the growth of physical AI. Do you envision a future where if you want to say you have a hybrid deployment where, let’s take a factory, for example, you have your humanoids running the Stereo Labs technology, and you have the large industrial robotics running the traditional Ouster Rev LiDARs. Do you envision a potential future like that happening?

Angus Pacala: Yeah. Right now I see the split where safety is involved, typically LiDAR is involved because it provides the Rev-8s are functionally safe devices. There’s a whole industry of legacy functionally safe LiDARs that are protecting people in manufacturing floors. We’re bringing that into the modern era with Rev-8. But safety, you can’t compromise on safety. And not compromising on safety means using functionally safe hardware. There’s just no way around that. So yes, I think that’s the split I see. There will be multimodal physical AI systems on a manufacturing floor. Big powerful robots doing metal bending and pressing and welding and advanced manufacturing robots navigating around the floor. These are big powerful machines. And then there could be these flexible humanoid robots and some will use LiDAR, others may not use LiDAR. They will definitely all use sensing of cameras or LiDAR in one way or another. Ouster is in a really good position to provide the best suite of those products of any company on Earth. I will also say that I think there’s a decent chance that flexible, safe robotic work cells are something that use LiDAR in droves. There’s a possibility that the industry will go towards making the factory ceiling kind of a smart, safe ceiling where there are functionally safe LiDARs looking down at all the robotic work cells and providing safety zones and monitoring as an additional layer over all of the activity going on below, so that the building itself kind of has the intelligence to trigger a machine to stop operating when a human approaches it. So there is a potential for really flexible work cell safety powered by the next generation of LiDARs like Rev-8.

Defense Applications

Grayson Brulte: There is a lot of opportunity there. The one market I want to touch on we haven’t touched yet is defense. You are a US-based company. The Department of Defense has a very large multi-billion dollar budget for automation. Do you see the defense business growing for you just because of where you are based and the ISO certifications that your company has?

Angus Pacala: Yeah, this is probably one of the topics I get asked the most about, and there is so much excitement on the retail investor side. I think a question about defense work comes up every quarter. Defense is not a monolith. There are many different applications in defense. One of the nearest term use cases is drone-based surveying. It is one of the use cases that makes the most sense to me because the technology works today and is fieldable today, and that is if you are going to build a business around something, you need the technology and the product to work in the near term. So we announced a year ago that we were the first Blue UAS certified LiDAR payload, which means that the DoD and other government agencies can use our LiDAR on these drone surveying payloads. So I think that the near-term opportunity into defense is very much in surveying payloads. There is a lot of activity in developing pseudo-autonomous or leader-follower or fully autonomous unmanned systems for the military. We happily support those efforts, and we actually have a long list of customers there.

Future of Ouster

Angus Pacala: But it is still developing in the same way that most physical AI applications are still developing. And the bar for these things working is incredibly high, both practically, you know this is an area where you can’t make mistakes. It can’t be a 1% field failure rate or error rate. You really need robust systems. And there’s also the making sure that the systems are behaving in a way that we are all comfortable with. I mean, this quickly goes in a direction of like everyone just immediately thinks “Terminator,” and we obviously can’t go and build that as a society. And so I think that the defense community is figuring out where automation even makes sense. And I’ve seen a lot of stuff that does make sense actually. So we have some great customers doing troop support vehicles. That’s actually probably the place I’m most bullish on is troop support vehicles where you have basically a robotic pack mule. Now historically, that was like a Boston Dynamics literal robotic pack mule, but now that’s moved to four-by-four, incredibly ruggedized four-by-four vehicles that are robotic and follow troops and carry their luggage, carry their gear, and make them far more efficient and able to focus on their task. So, all that is to say Ouster has a bunch of defense work that we do, but I think it’ll take a couple more years for it to be a business that really impacts the growth trajectory of Ouster outside of the drone surveying work that we are in a clear kind of pole position for.

Grayson Brulte: Well, it is clear that you and your customers are building the future, and I guess putting this entire conversation together, what is the future of Ouster?

Angus Pacala: The future of Ouster is to lead the world, be the world-leading platform for unified sensing and perception. I want Ouster to be the name that when you know that you are interacting with a robotic system that is using Ouster technology, that you immediately feel comfortable. Because we’ve established such a track record of safety that you know that this system, whether it is a car or a humanoid or a traffic light or a drone or a small delivery vehicle delivering you a package or a forklift if you are working in a manufacturing floor around forklifts or AMRs or robotic arms. I want Ouster to be the established brand that just gives you peace of mind because we’ve built the safest, most reliable systems in the world. I think there is a lot, it’s scary. A world where these systems aren’t safe or aren’t built right is a scary world. And Ouster, I want to be the company that does it all right. And I think that’ll resonate with the engineers in the audience. Largely our strategy at Ouster is to build things that truly work, do great engineering and not cut any corners and solve the hard problems for our customers. And a lot of that now comes down to building these incredibly safe and secure systems. So, in a nutshell, great engineering is the strategy and business model of Ouster, but the vision is unified sensing and perception for all of physical AI. Some company has to do it. We want to be the company that does it and succeed because we’ve done it so well that you really can’t not use our products because it’s the safe and best way to do it.

Grayson Brulte: I will summarize this way, Ouster is building trust by doing it the right way. The future is bright, the future is autonomous, the future is Ouster. Angus, thank you so much for coming on The Road to Autonomy and sir, we are not going to let 14, 15 months go by before we have you back on again.

Angus Pacala: That is a deal. This is always fun, so happy to come back.