[00:00:00] Speaker A: We are using vision to look at the environment around us and constantly compare what we see to what we've seen before. With that, you can constantly recalculate your position in space and provide the position in space without accumulating noise.
[00:00:21] Speaker B: Hello, everyone, and welcome to the Robot Industry podcast. We're glad you're here and thank you for subscribing. My guest for this episode is Amir Busani from Argo Robotics. Amir, welcome to the podcast.
[00:00:32] Speaker A: Hey, Jim, it's a pleasure to be here with you. Yes.
[00:00:35] Speaker B: So listen, if you wouldn't mind, maybe tell our audience a little bit about you and kind of how you got into the robotics industry.
[00:00:42] Speaker A: So, yes, I would say that was meant to be. From my early days as a child, I admired robotics. So some kids likes animals, I like robotics and machines. So it was really meant to be. My degree was in computer engineering, and I spent many years in different positions and management positions. And when I got the first chance to start looking at computer vision and robotics, I was immediately attracted to that. And it was obvious to me that that's really what I want to do. My previous role was the manager of Intel Real Sense Advanced Technology Group. So I got the opportunity to work with many leading robotic companies to assist them in building the new robot and to see how vision and new 3D vision technologies can really elevate and assist robotics in solving problems that they weren't able to solve before.
[00:01:32] Speaker B: And so what made you decide to kind of focus your energies on vision and autonomy?
[00:01:38] Speaker A: So that was back in 2018, and it was a point in time where everyone saw the huge progress in autonomous vehicles and how computer vision cameras and AI are now enabling things that seems to be impossible before. But on the other hand, we saw the big challenges in autonomous vehicles safety and the time it will take to get an autonomous vehicle to the safety level that is really needed to deploy. While in parallel, we looked at robotics and taking a smaller machine to operate in the environments that are not that dangerous and capable. On the other hand, the gap between what robotics can do and what robotics are doing today is really exciting. I mean, there's endless opportunities and use cases for robotics, and we're seeing only the tip of the iceberg when it comes to robotics. So it was really a revolution that I want to be part of.
[00:02:35] Speaker B: Oh, that's great, but maybe for the audience, because we've got lots and lots of people all around the world. Could you tell us what Slam stands for? And especially visual slam?
[00:02:47] Speaker A: Yes. So in robotics, the first challenge is to know exactly where the machine is. So you want to know your exact position and to understand how you move in space to solve that. There are many different techniques, but the common one is called simultaneous localization, and mapping. The idea is that you use your movement in order to calculate the 3D positions in space around you, while in parallel calculating your relative pose, coordinated XYZ position and also the angle in each one of that of your movement. Visual Slam is doing that with vision. So for a mobile platform, you can take a picture with your camera. You move a little bit with a robot and then you take another picture. And then on the image frame itself, you can compare the relative changes between the two frames and from that you can generate some equations of what changed. With enough points, you can solve enough equations to solve simultaneously the fixed degree of freedom position of the camera, so the coordinate of the camera and the angle, while also detecting the 3D positions in space around you. So that's what generate the map together it creates a map and the path of the robot within that map.
[00:03:59] Speaker B: And thanks for clearing that up. And can we also talk a little bit about AI and how kind of generative AI comes into the picture?
[00:04:07] Speaker A: Yes. So first maybe a few words about the challenge with the Visual Slam. So visual slam generates error. It accumulates a little bit of error between every frame, but that error accumulates over time. So alone that technology has its limitation when it comes to the accuracy it can achieve over time. For that reason, vision and learning the environment can assist. We are using vision to look at the environment around us. It's kind of like how humans do it and constantly compare what we see to what we've seen before. With that, you can constantly recalculate your position in space and provide the position in space without accumulating noise. Now, that's only really the first step when it comes to AI in robotics, because beyond that, you would want robotics to have high level of understanding or human level understanding of space, so they can understand the geometry of space around them, so they can detect different types of objects and to know how to behave correctly next to these different objects. When you're thinking about a forklift driving autonomously in the warehouse, it can move very close to the shelves, but if there's a person in front of it, you would expect it to stop and take some safety distance. That level of behavior can only be achieved with a machine that has the AI capabilities and the understanding of how to behave correctly in space.
[00:05:38] Speaker B: No, that's very clever and thank you. That's very clear and wanted to kind of understand a little bit about how big this industry is because you are creating hardware and software and maybe firmware for the industry, right?
[00:05:52] Speaker A: Yes. So the robotic industry is growing very fast. So the new areas in robotics are growing almost 30% year over year. It's actually impressive that even the most challenging years, like during COVID and even when the global economy is impacted robotics is constantly growing. On the other hand, there are endless new opportunities in robotics. So there are only few use cases that are currently really reaching the volume that they can reach. And it's clear that robotics is going to continue to grow. It is estimated to reach about $40 billion within the next couple of years. So, huge opportunity when it comes to robotics. On the other hand, it is clear that the enabler of new types of robots and the continuous growth in robotics will require smarter robots. We call it intelligent autonomy. You would want your machines to be intelligent so they can be autonomous, even in more challenging environment. And that's really an enabler to the future of robotics.
[00:06:56] Speaker B: You know what I was going to ask you, that what that question was anyway. So that's perfect. And in terms of dollar terms, do you have kind of some feeling for how big your addressable market is for this industry?
[00:07:09] Speaker A: Yes. So the estimated part of perception within robotics is estimated that between 15 and 30% of the overall bill of material of robotics. It makes sense when you think about that, because at the end of the day, a robot is usually a mechanical design with wheels and motors, and it's really the brains and the sensing part that makes it a real robot. And it's a big part of the overall robotic system. We also believe that it's going to be a bigger part as we go where the eyes and the brain of the system is going to be a more fundamental and differentiating part of robotics. It is currently estimated as about $12 billion for the perception for robotics spread across different industries and different use cases.
[00:07:55] Speaker B: Of course. Thank you for that.
Can you tell us a little bit about your team? Actually, I went to your website earlier today and I had a look and I looked at your job postings and was impressed by the size of your team.
[00:08:08] Speaker A: Yes. So team in Algo is focusing on providing perception, and most of our team members are experts in the domain of computer vision, AI, sensor fusion, and embedded systems designed for robotics.
We have a large team, over 40 experts in these domains, with a lot of not only the academic know how, but also the real world understanding of the applications and the difference between a walking technology into a walking product and also the understanding of scaling. So what it means to take a product and scale it into the mass market with volumes and with all the applications of manufacturing deploying in many different use cases and many different environments in many cases. That's really what makes the difference between good technology and a good product for different use case, but especially for industrial applications where the quality is all that matters.
[00:09:12] Speaker B: Of course. So did we explain this yet what a perception engine is?
[00:09:16] Speaker A: I'll be happy to do that. So when we talk about perception. We're thinking about the information a mobile machine needs in order to move in space, the understanding of space that it sees. So, in a way, the perception engine enables a machine to see and understand its surrounding so it can move autonomously from one point to another. The main challenges in terms of perception are, first, understanding the position of the robot in space, just knowing where you are at any given moment.
Then there's the understanding of the geometry of space. So what's the surface I'm driving? Where are the obstacles? What's the distance and height of different obstacles? And from that, the robot needs to understand, what's the maneuverable area? Where can I drive? How should I move between the different objects? Some of them are static, some are moving. And how to calculate my path and to get to my destination. The higher level of perception, which we call human level perception, is to understand how to behave in this environment. So to understand the different environments, different objects and their vector of movements, to get the level of understanding that you can move intelligently, even in challenging environments, just like humans do.
[00:10:32] Speaker B: The industry is so big with so many different markets. So are you kind of concentrating on one particular area or is it kind of everywhere?
[00:10:41] Speaker A: So you're on to a great point. That was our challenge from the beginning. I mean, it's just so tempting to get into all these different use cases, and all of them are exciting. Agriculture and logistics, industrial service robots. What's not the challenge for a company, and for me as the CEO, is to make sure that we take the right focus at every given point along the way. There isn't a week when I don't get a call from a new robotic company with an amazing use case. Really, some of them are exciting, but we have to make sure that we do it step by step and that we address the market correctly. We've actually started with outdoor robotics, in a way solving a broader problem than indoor robotics. And that helped us in making a solution that is very robust. I constantly get a question from warehouse owner. So what do you do when the window is open? We can operate in direct sunlight. We know how to solve that. Today, the focus is both outdoor and also warehouse environments, logistics, ecommerce, industrial automation, where most of the robotic market and most of the deployments are today.
[00:11:54] Speaker B: I like your model. It enables a company to do the hardware of the robot and not spend years developing the hard challenge of navigation. And you kind of have a cost optimized solution, a standard and a premium offering. Can you explain the differences? And maybe because people are there's no visuals here. What does it look like? It's a lens attached to a circuit board, right?
[00:12:17] Speaker A: Yes. So our base system is designed with just a simple camera, and we designed the camera system to be very simple, very easy to integrate. It's a simple small camera that you integrate into your robot. And the rest is smart software algorithms running on affordable compute. When we looked at the robotic market, we realized that the breakthrough solution for that market should provide the highest level capabilities, so high level perception, high quality, but it has to come at a solution that can fit the bill of material limitations of every mobile robot. So it has to fit a cost that can be affordable enough to deploy en masses. And that's why our base solution is based on very affordable components, in a way the lowest cost components you can think of for a robot that can fit camera and arm level compute platform. And on top of that, we can add additional sensors, LiDAR, 3D cameras, GPS and many others. So we can also support more complex environment and machines that can add some additional sensors and can bear the extra cost.
[00:13:31] Speaker B: You're located in Israel with an office as well in Boston, and how many people do you have in each office? And do you see the US growing?
[00:13:38] Speaker A: Yes, certainly. So we actually started a company in Boston.
We see Boston as the mecca for robotics, with companies like Kiva System, Irobot, Boston Dynamics and so many other amazing companies in that domain. It's really the best place to grow a robotic company. We're leveraging a lot of the talents we have in Israel and we now have offices also in Europe addressing the European market.
We are growing in the US, covering more and more companies within the US where the innovation and the best business and some of the most innovative companies are.
[00:14:17] Speaker B: And can you tell us, with so many companies today, data is a really important part of the company. Do you collect this data? What do you do with data?
[00:14:30] Speaker A: So yes, data is crucial for every application using Vision and AI, we collect the data and we also use the data to constantly improve our system. So every new environment we see and every new challenge that we encounter enables us to improve the system and to make it more robust to changes. And our horizontal solution, the fact that we run in very different environments allows us to cope with different types of challenges and to make the overall system as better. So we see indoor challenges and outdoor challenges, challenges with big machines like forklift that drives quite fast and with machines operating with many people around it, and the challenges of dynamic environment. All these data sets and all that data that we collect helps us improve our system so it can cope with future challenges when we get into new sites.
[00:15:30] Speaker B: And you must have a lot of different partners, obviously you've got like funding partners and you've got technology partners. Is that an important part of your startup?
[00:15:38] Speaker A: Certainly, we believe in partnership partners we have and the companies we work with are I would say it's almost like marriage. Because when you select your partners right, that's who you are going to win or lose with. Our partners are from different domains and different types of companies. We have industry leaders and market leaders in certain domains and on the other hand, we like the disruptive companies with new products and innovation and the fast pace of bringing new products to market. We're working with companies in Europe and the US and in Asia, also leveraging the progress done in different geographies. It's an important part of growing a robotic company. And from the beginning we knew that we have to look at that globally and to work with companies all over the world in order to be able to provide the solution that can drive the next generation of robotics.
[00:16:29] Speaker B: So tell us a little bit about some of the and we don't really need to know the customer names, but when you get these calls, who are they? Are they just everybody like all these robot companies?
[00:16:40] Speaker A: Our customer profile is so varied. It could be a huge multibillion dollar company with a team of 50 working on similar technology that realize that we have a breakthrough technology and that they want to collaborate with us. Or it could be a CEO that had the idea of building the company and now we realize that if he gets the perception technology from us, he can focus on his application and get the product to market three years earlier of his plan. So very varied in that aspect. And what I do see a lot in the robotics industry is that many of the leading companies, they have the passion for robotics. So in many cases, it's not only about the business, it's a passion to build new robots, to try new technologies and to drive innovation.
[00:17:27] Speaker B: You must get some customers too that are in trouble, right? We've been working on this for months and months or years and years and we haven't solved it. Can you help us? You must get those calls too, I imagine.
[00:17:38] Speaker A: Yeah, that type of cases where you talk to the CEO and you can see how happy is to see that he found a partner that can help him solve that part of his problem so he can move forward. Perception is a huge challenge and I just don't think it makes sense that most of robotic companies are still working on that and trying to solve that in house with simple sensors, that was doable. So with line following technologies or simple proximity sensors, that was doable. But when thinking about perception using cameras and AI, it's a huge problem to solve. It requires a huge team and many years of development. It just doesn't make sense for every company to try to solve it on its own.
[00:18:24] Speaker B: Again and again I see the development process a bit like a bubble, right? The bubble kind of moves through the startup. And like you say, some bubbles are just too big.
What do you see as the future of perception and autonomy?
[00:18:40] Speaker A: So I think, you know what, it's sometimes the kids that see it the best. But when you think about robotics, if you'll ask a kid what he thinks about robotics, what a robot should do, that's what should happen, right? They would imagine a smart robot that can do all these things and that can take smart decision. But in reality, today, when you look at most of the robots, if you look at an industrial robot, at best it can follow a line in the warehouse from one point to another carrying a box, but not much more than that. So there's so much more that can be done. I think the next big revolution, and it will be driven by generative AI. And the new generation of AI will be to switch from current systems that though they are autonomous, their programming process is all manual. It's all coded manually to do one very specific task. Every point in the past is manually programmed into smart machines that can take smart decision. You would want to be able to tell your robot, okay, robot, here's my Warehouse. Run around a little bit, learn it, and come back to me when you know how the Warehouse looks. And then I'll tell you where to go, right?
Two years ago, it sounds like fan fiction, but today, with generative AI, you see it coming, and it will come in the next couple of years.
[00:20:01] Speaker B: Now, it's going to be exciting time. Amir, have we forgot to talk about anything today in the podcast?
[00:20:08] Speaker A: There are so many topics to discuss. When it comes to robotics, I think the most exciting thing is how the market is going to go and to be at the point of time that you know that in two years, three years, there will be so many new applications and so many new products. Just exciting.
[00:20:26] Speaker B: And I imagine you have all these startups coming, and you get to think with them and ideate with them. And it is a very exciting part that you're in.
Amir, when you're not in knee deep in software and hardware and growing your team, what do you like to do for fun?
[00:20:40] Speaker A: So though we do robotics, the thing I like to do the most is to walk by my own. So hiking is something I like to do, which is funny, because for many of our customers, we assist their employees to walk less where robots can carry goods instead of them. But my actually hobby would be to go hiking and traveling.
If there's a mountain around, that's even better, that's great.
[00:21:07] Speaker B: We actually just did in this summer, we were just in Newfoundland, Canada, and did a lot of hiking. So if you ever get a chance to go up to Grossmorn Park in Newfoundland, it's amazing. And how can people find out more about Argo Robotics.
[00:21:21] Speaker A: So of course, you're all welcome to enter our website and to view it, you're welcome to follow us on LinkedIn and get the latest updates, our podcast, and the conferences that we join. We're happy to talk to anyone in the robotics space. Anyone that is interested is very welcome to contact us and be part of this revolution.
[00:21:46] Speaker B: Amir, thanks for joining me today.
[00:21:48] Speaker A: Thanks a lot, Rosan.
[00:21:50] Speaker B: Our sponsor for this episode is Earhart Automation Systems. Earhart builds and commissions turnkey solutions for their worldwide clients. With over 80 years of precision manufacturing, they understand the complex world of robotics, automated manufacturing and project management, delivering world class custom automation on time and on budget. Contact one of their sales engineers to see what Earhart can build for you and their
[email protected]. Earhart is spelled, ehrhardt? And I like to acknowledge a three. The association for Advancing Automation. They are the leading automation trade association for robotics, vision and Imaging, motion Control and Motors, and the Industrial Artificial Intelligence Technologies. Visit Automate.com to learn more. And if you'd like to get in touch with us at the Robot Industry podcast, you can find me, Jim Beretta on LinkedIn. Today's podcast was produced by Customer Attraction Industrial Marketing. And I'd like to recognize my nephew Chris Gray for the music, jeffrey Bremner for audio production, my business partner, Janet, and our sponsors, Earhart, Automate Nation Systems.
