Speaker 0 00:00:00 There are no collaborative robots. What there are collaborative applications and those can lead to great productivity and cycle time improvements for manufacturers. What we do at Vail robotics is we make it possible to have collaborative applications that involve standard industrial robot.
Speaker 1 00:00:20 Welcome everyone to the robot industry podcast and hello to our listeners in St. Paul and St. Charles. My name is Jim burrata and I'm your host. Mike guest today is Patrick civil Beryl. Patrick is the co-founder and CEO of Veo robotics, a company that brings advanced computer vision and 3d sensing to industrial robots, allowing them to work side-by-side with humans in manufacturing processes. Patrick has more than 25 years of experience in computer vision, robotics and industrial automation prior to founding Veo robotics. Patrick was the first entrepreneur in residence at Siemens venture capital. Focusing on the creation of the vision behind Veo. He also served as the VP of new growth platforms for Avery Dennison, building a new business line in RFID based IOT products. He was president of rethink robotics, creators of collaborative manufacturing, robots, and founded and led the computer vision startup in television to its acquisition by Tyco international, originally trained as a computer scientist. Patrick holds a PhD, an Ms, and a BS in computer science from the Massachusetts Institute of technology. Welcome to the podcast, Patrick. Great
Speaker 0 00:01:42 To be here, Jim. Thanks very much.
Speaker 1 00:01:44 Well, we said in our intro there that you've got a lot of friends who've been on the podcast, so we're very excited to have you to tell us a little bit about VO. So for those people out there, listening all around the world who don't know about VO, can you explain what you do and what problems you're solving?
Speaker 0 00:02:01 Sure. Uh, we make a 3d safeguarding solution for industrial robots, standard industrial robots from FANUC U Scala, ABB and KUKA. And what our system does is it makes it really easy to add human robot interaction to a work cell that involves a standard industrial robot. Our system uses 3d time of flight sensors, uh, that we provide, and it's a safe system. Uh, it will be certified to ISO 13, eight 49. So that's the same standard that you'd use for any, uh, electronic, uh, safeguarding electro-optical safeguarding equipment, uh, such as a light curtain or an area scanner, but it provides much more, a much more comprehensive solution, a solution that allows, uh, 3d vision and semantic understanding of what's going on the work cell that makes it, uh, much easier to install. We reduced the total cost and complexity of human robot interaction.
Speaker 2 00:03:08 And so how did you get started? I know in the bio, we kind of mentioned kind of where you got started, but how did you get started,
Speaker 0 00:03:14 Uh, at, at Vail robotics? Yes. Uh, so, uh, you know, when I was, uh, many years ago when I was the first president at rethink robotics, I, uh, wrote, uh, the business plan, uh, which was, uh, really the first, uh, business plan for a venture backed collaborative robot startup. And, uh, we went out and talked to many, many different manufacturers. Uh, some of whom are our customers today. Our approach was always to go down the manufacturing line together with manufacturing, engineers, and talk about problems that they had and solutions they were looking for. Uh, and, and generally brainstorm, uh, about what the needs were rethink robotics, of course, uh, went in its own direction. And I left the company actually after a year and 10 months, uh, because they were very focused on building a, a humanoid robot, uh, which I didn't see as responsive to customer needs, but we've seen power enforced, limited collaborative robots be very successful in industry, uh, all the same when going down manufacturing lines at, uh, for example, durable goods manufacturers like automotive OEMs, uh, or tier one components makers, or, or, uh, major appliances or, uh, you name it often that power enforced limitation that makes, uh, the current generation of collaborative robots, safer human interaction.
Speaker 0 00:04:58 Also post real limitations for those manufacturers, uh, power enforced limited robot can't reach as far as a standard industrial robot. I can't move as fast. Can't be as strong and sometimes not as repeatable. So, uh, I was in the BMW factory in Spartanburg when I had the idea for Vail robotics. And that was back in 2009. I knew that the technology of the time, uh, the, the computational power and the, uh, sensing, uh, capabilities of the time, wouldn't be sufficient to allow the implementation of it. But I always had it in the back of my head and, uh, Clara VU. My co-founder was a contract product manager I was working with and visiting customers with at the time. And so when, uh, I saw that the technology had the base technologies, we would build our solution on, had moved ahead sufficiently. I joined Zeeman's as an entrepreneur in residence, and I called Clara up and I said, Hey, uh, you know, it's time to do this thing.
Speaker 0 00:06:04 So, uh, that's pretty much how we got started. Uh, we have another, co-founder our VP engineering, Scott Denenberg, uh, who I had also, uh, met, uh, years before. And, uh, we put together a proof of concept, an early prototype and, uh, set out to get a safety certification, a functional safety certification. So, uh, uh, along the way we've raised, uh, almost $40 million with top drawer investors like Google ventures and lox capital, uh, Zeeman's of course has been great to work with. And, uh, they are an investor, uh, but you know, our latest round was led by a European investor, Alvin intelligence capital, and, uh, we are on the verge of receiving our functional safety certificate.
Speaker 2 00:06:53 Oh, that's wonderful news. And at the, Oh, your robot agnostic and that's correct.
Speaker 0 00:06:59 That's right. It takes us, uh, about a day and a half of a single engineer's time to add a, a particular robot model that's compatible with one of the controllers from the four robot manufacturers we work with. So we have, for example, uh, added those a major construction equipment maker had a, an early version of a new FANUC robot. We were able to add those without ever having one of those robots at our site. And, uh, we can also bring on other robot, uh, controllers, uh, new controllers, uh, from other, uh, manufacturers in, in pretty quick order that that's, uh, a longer, uh, process, but it takes a few weeks to, uh, uh, get a new controller model up and running. And, uh, then only a day and a half for the individual model.
Speaker 2 00:07:56 And I assume once you've got that controller or that robot kind of built into the system, then it's just a copy paste, correct.
Speaker 0 00:08:03 Uh, that that's right. Um, of course, manufacturing engineers do, uh, very, uh, complicated work in work cell design. And, you know, this, this gets back to what we were saying at the very beginning of the show, uh, where, um, there are no, uh, collaborative robots, there are collaborative applications. And so a manufacturing engineer has to decide, you know, what is the human doing? What is the robot doing? What is, you know, what is the other equipment in the work cell? And, you know, how do I, uh, make sure that this gets done in the cycle time budget that I have, uh, how do I make sure that the process step is this done with high quality and high repeatability? Uh, but, um, what we've sought to do for manufacturing engineers and our assistance integrator, uh, partners is, uh, provide a product that is really easy to, uh, put in as the comprehensive safety solution for the work cell and really easy to integrate. Typically we're talking about a day or so to get the, uh, system connected and configured for working in a, uh, in a, uh, safeguarding application.
Speaker 2 00:09:19 Thank you for that. What do you have a best like bulls-eye customer or so obviously you mentioned automotive, but is there any other particularly focused industry targets that you're going after?
Speaker 0 00:09:30 You know, uh, we, we of course have done a bunch of work with, uh, both automotive OEMs and tier one automotive components manufacturers because, uh, they are, um, so sophisticated in their use of robots. And traditionally, as you know, they, uh, counted for the biggest chunk of new robots sales, uh, in the industrial robot industry. Uh, that said, uh, automotive makes up probably about a third of our customer set. We also work with consumer packaged goods, uh, makers. We work, uh, we have aerospace customers, um, people earlier in, uh, uh, you know, kind of the component supply chain, uh, people who might do something like a metal casting or, uh, the provision of other, uh, uh, components into, uh, into products. Um, so, uh, we will work directly, uh, with a manufacturer or through a systems integrator, uh, being a good partner to systems integrators is, is really one of the keys to success in industrial automation. And so we've launched a certified integrator program and, uh, we hope soon to be able to announce our first certified integrator, uh, which is, uh, uh, a really superb integrator. And, uh, we're looking forward to, uh, doing an announcement together.
Speaker 2 00:10:58 Oh, that's great. I agree with you that integration and having good integration partners is key to success. What hardware and software then comes with VO. And do you see much, uh, obviously a probably evolve, uh, kind of Greenfield sites, but do you see much work in retrofit?
Speaker 0 00:11:14 Uh, so the heart, let me answer the first part of your question first, um, we've, uh, needed to develop both hardware and software. Although most of our engineering team is actually on the software side, just because of, uh, the intelligence of the system and, uh, the amount of computer vision computation it has to do. And, uh, the sophistication of those algorithms for, uh, you know, uh, dynamic modeling of what the robot is doing and controls and so on. But in terms of the hardware, we, we have to run that software on, uh, very high performance hardware. So we make, uh, a device called the free move engine and what it has inside. It is two entirely separate dual process or motherboards that, uh, uh, are doing a tremendous amount of computation, the very high end, uh, Intel processors on those boards, as well as an external, uh, well inside the box, but, uh, off those motherboards, there's another motherboard, uh, that, uh, we built in design that has a safety processor on it.
Speaker 0 00:12:25 And that safety processor is checking the results of those two computations, uh, uh, being executed independently on those motherboards against each other. So we have dual channel redundancy inside that free move engine, which is what we call that box. And then we also develop dual channel sensors. Uh, and those dual channel sensors are IP 65. Uh, they, um, have two separate imaging elements and optical paths. And then there's a, uh, uh, safety rated FPGA that is, uh, comparing pixel by pixel, uh, the sensing on those, uh, two separate optical paths, uh, to each other. You can attach up to eight of those sensors to one of our free move engines, which gives us the ability to monitor work cells as large as, uh, eight by eight meters. And, um, get a, uh, a full 3d view of what is going on in that works out, uh, whether it's, uh, people moving around, uh, we recognize the robot it's its end effector, uh, the work pieces it's carrying, uh, other hazards, uh, present in the cell, uh, and anything that is large enough to potentially be a human that we don't have a CAD model that's been loaded into our system for, we're going to treat that as a human, and we're going to control the robot to maintain a protective separation distance, uh, and bring the robot to a safe stop before a human can reach it.
Speaker 1 00:14:04 And that's good to know. I was going to ask you about the, you know, the, the, the system differentiating between, uh, somebody with legs and maybe some tooling that I have to keep away from. Yeah,
Speaker 0 00:14:14 There's a, you know, there, there's a tendency, I think among say folks who are enthusiastic in computer vision and robotics, uh, when they're getting their education, you know, imagine a university, a doctoral student and their advisor, and they say, well, we're going to use machine learning to recognize people. And, uh, then we'll be able to identify people, uh, in a work cell and we'll be able to do robot controls that way, the difficulty with approaches like that is that machine learning, uh, really is a kind of statistical pattern matching approach that, uh, depends, uh, it's success depends very much on the size of the training set and, you know, it's good enough so that, you know, Google or Facebook and try to find, uh, pictures, uh, with your friends faces in them, but sometimes they'll get it wrong. Well, we can never get it wrong.
Speaker 0 00:15:14 So our approach is much more deterministic. We don't use machine learning in, uh, any part of our safety function precisely because a statistical approach, uh, isn't good enough for us. We have to know, and we have to, uh, therefore, you know, be absolutely certain that the robot can move safely before we allow it to move at all. So anything you do to our system, you run a forklift into one of the sensors, or, you know, you power off the box or, uh, destroy the box in some fashion. Uh, we will always, uh, fail to a safe state. And that's what a lot of our, uh, FMI and FMI DA's are about and why we've had to, you know, build up from the component level, that safety case. And it's taken us, took us three and a half years to, uh, get our FMAs and FM EDS, uh, done in the course of our hardware design. You know, you, you, uh, as Clara says, sometimes that, you know, rather famously around five years ago, um, Mark Zuckerberg said that, uh, his philosophy at Facebook was move fast and break things. And, uh, Clara, our CTO says you can't move fast and break things. If those things are people, absolutely safety is absolutely crucial from our point of view. And, uh, it's, uh, a deep ethical responsibility as, as well as a business imperative.
Speaker 3 00:16:45 Of course, if I was to walk by a VO powered robot style, what would it look like? What, how would it be different? So, um,
Speaker 0 00:16:53 What you would see would be, uh, mounted probably overhead, although you have full flexibility in where you place these sensors, but probably somewhere overhead, either at the periphery or directly overhead in, in the work cell, there would be these green boxes and there maybe about, uh, 10 inches by six inches, uh, and those houses are sensors. And of course they say fail robotics on them and their cables coming down from those, their power over ethernet. Uh, and they go into our controller, our free move engine, uh, which is a separate box about the size of a, uh, a robot controller, a small robot controller, and then their cables that go from that into the robot controller. And, uh, that's how we slow and stop the robot. And that would, uh, be what all you see you, you asked the question earlier about whether it's Greenfield or retrofit, and the answer is both, uh, you can add this system, uh, as a retrofit to an existing cell.
Speaker 0 00:18:02 And, um, you know, it's, it's really a matter of, uh, placing sensors correctly. And, um, uh, we have, uh, intelligent software that allows you to, uh, register those sensors. You don't have to do like a bunch of, you know, calibration and, and, and so on. Uh, we, we can use the robot as a fiduciary to figure out where the sensors are located in, in 3d space. And again, that's, that's key to making it really easy to set up this system, but, uh, you, you might not even notice, uh, the presence of, uh, the, the veil system because, uh, um, really, uh, we try to take a light touch on the robot controller. We don't replace it, we supplement it, and our sensors are, uh, built to be really easy to put in flexible locations. And, uh, what you would notice is that if you approach that robot, it would, uh, slow or go into a safe state when you were close. And if you walked away, it would, uh, resume operation automatically.
Speaker 2 00:19:13 I think that's always the best way to tell. Yeah, that's right. So who is your customer? Is it health and safety? Is it the application engineer or the plant manager? I mean, on the, on the manufacturing level, of course at the integrator level, it's probably someone else. So,
Speaker 0 00:19:29 You know, as, I don't need to tell you, since you've been in the industry for quite some time, there are manufacturers who insist on buying direct and they have the, uh, internal capability in their advanced manufacturing group to assess, uh, and, and build applications. And, and of course we work with them. We, uh, also, uh, um, have systems integrators as, as customers. And we see them as really key to our strategy, uh, and, um, you know, they, uh, can resell, uh, the, the free move system. And, uh, we support them with training and, uh, marketing collateral, but you know what I'll often say, what explaining this to somebody who doesn't come from the industry is our customer is not a washing machine manufacturer or, uh, uh, a, uh, a car manufacturer. Our customer is the person inside that entity or outside that entity who makes the factory, our customers make factories.
Speaker 0 00:20:37 Uh, they, they make lines, they make cells. Um, and, and so that's who, uh, whose needs we're really, uh, seeking to address whose problems we're really seeking to solve. And so, uh, health and safety of course, are in the room on any solution that involves human machine interaction. Uh, and, and we absolutely, uh, uh, spend time with those folks. And it's, it's one of the key reasons why even for sales within the United States, where it's not mandated by law, we've, uh, gone for a functional safety certification from TUV. Um, because, uh, we want to answer that, that question right up front, is this a safe solution? And, um, in Europe, uh, of course it is a requirement under the law to, uh, have that functional safety certification before you install that equipment in a manufacturing facility. And so, uh, it's been part of the plan from the very beginning.
Speaker 2 00:21:44 That's great. And how are you getting invited into a plant or a facility? Are you, is it a lot of word of mouth or, uh, as some of the other guests on, uh, on the robot industry podcasts, they've actually recommended and talked about you, so I'm wondering how does that work?
Speaker 0 00:21:59 You know, um, it's a combination of things. Um, it, it, it, isn't always the case that, um, startups have experience in the industry they serve, uh, in, in the market that they serve. Then if you think about it, um, really, uh, if you read, say tech crunch or wired, or one of those, uh, uh, places where startups are written about, um, a lot of what they're talking about is consumer products and services. Uh, so, uh, it makes sense, you know, if you look at the companies with the highest valuations, uh, that are trading in the public markets today, they are often very consumer facing companies. Our approach was to try to be really good citizens in the industry from the beginning. So, uh, you know, we got a participation in the standards committees, uh, tried hard to make a good contribution there. Uh, we didn't know a number of, uh, people in the industry already.
Speaker 0 00:23:07 And, uh, so, uh, we were able, you know, I was talking to, uh, some customers in consumer packaged goods about this company before it was incorporated. Um, so, uh, you know, just making sure that we're really responsive to customer needs and to a great extent it's been, it's been word of mouth. We actually did do, uh, we did first show our own hardware running our own software at the fab tech trade show, uh, in November of, uh, of 2019. Uh, so that's about a year ago, more than a year ago now. And through all those, you know, this is a highly referential group of people. And so by making it very clear that that what we're seeking to do is, uh, serve and, and provide something that is a value, as opposed to, you know, what people like to talk about is disrupting markets and so on.
Speaker 0 00:24:07 Uh, you know, that's not us, uh, you know, our goal. We, we see making things as so fundamental to, to human prosperity and fulfillment and success, and, and we love factories. And so our approach from the very beginning has been, let's make, uh, something that, that, that makes that more productive, makes it better. And, and we feel very, uh, very much on the side of the angels with that, you know, uh, if we can, if we can make it, you know, less expensive and, and more responsive to consumer needs to build a new class of refrigerator well, that's, that's good for everyone. Um, uh, similarly with, uh, with, uh, transportation, uh, anything you want to talk about along the way, people tend to forget about factories, um, and, uh, they just assume that things come from stores or, or whatever. Uh, but, uh, you know, it's, it's so crucial to, to, to all of civilization, um, and, and we love factories. So, uh, it works out.
Speaker 2 00:25:17 I'm a big fan of factories too, and I think it's exciting that you're making a product that makes, uh, makes it safer for people and equipment to, to work together. What's next for Vail? Do you, where do you see the technology going
Speaker 0 00:25:29 Very much? Uh, again, customer driven. One of the things we hear from, for example, an aerospace customer is they would like us to cover larger volumes. So I mentioned, you know, eight by eight meter work cells. Um, you know, if, if they're doing something like, say a manufacturing, the wing of an aircraft, they'd like to be able to cover that entire wing, and these wings can be very large, but we also, uh, think about, uh, you know, more complex applications, so free move, uh, one controls, one robot at a time. Uh, we'd like to make it so that, uh, we can handle multiple robots, uh, robots that move, for example, on rails over, uh, uh, say 30 meters. So we'd like to expand the volume that we can safeguard, uh, at once. And, you know, we have, uh, you know, uh, some advanced work on that we've already, uh, filed some patents on it.
Speaker 0 00:26:30 And, uh, you know, another thing that has happened, for example, in logistics, but not really so much in manufacturing today is, uh, the use of AGVs. If you look at, uh, manufacturing, the processes are far more complex, then, uh, the prophecies in say e-commerce logistics. And so the use of AGVs has been a little bit more complex, uh, harder to integrate and by, uh, being able to, uh, monitor large spaces, uh, including the ones that both people and AGVs are moving through, we can, we can help there. Uh, we'd like free move to be something that, uh, you can just depend on as a manufacturing engineer, when you're designing a process staff and designing a work cell, uh, you know, it's, it's, it's there, it's present just like you got four 80 volt, three phase electricity, and you got 120 SCFM, uh, compressed air, you know, you've got human machine safe interaction available to you.
Speaker 1 00:27:38 Thank you for that. We've got some great challenges ahead of you, and I can hardly wait to see it in action at the next trade show when we start going back to trade shows again.
Speaker 0 00:27:48 Yeah. Well, it's been a, it's been a bit of a, a problem for the industry that, um, uh, uh, of course the pandemic has, has been, uh, a global tragedy, but particularly in our industry where you really want to be able to see things in action and, and really believe them, uh, and, and understand them. It's, uh, it's a shame that, um, uh, we haven't been able to do that, but, uh, we, we certainly look forward to being able to do it again.
Speaker 1 00:28:18 Yep. I'm going to be first in line to get my shot. Hey, Patrick, how can people find you and find out more about Vail?
Speaker 0 00:28:25 So our website is, uh, uh, Vail bot.com V E O B O t.com. Uh, I can be
[email protected], uh, and, um, we're, uh, we're always delighted to, uh, talk to anyone in the industry. So, um, uh, please, uh, anyone listening feel free to reach out. Uh, we are adding integrators to our certified integrator program, uh, and, uh, they're, they're great incentives and support. So, uh, we're, we're happy to talk to anyone in the industry and, uh, see if, uh, we can, we can be of help.
Speaker 1 00:29:09 I'd like to thank our sponsor for this episode. So thanks goes out to new scale robotics who design and build the Q span, automated small part measurement systems that automates digital caliper measurements, data logging, and part handling KeySpan picks, measures records, and places, the Q span system. It's the new standard for QC solution for high mix, small batch manufacturing, and works with universal robots, wine of collaborative robots. So it's easy and safe to deploy in your QC lab. I'd also like to thank and acknowledge our partner 80. The association for advancing automation. 83 is the umbrella association for RIA AIA MCMA and eighty-three Mexico. These four associations combined represent almost 1300 automation manufacturers, components, suppliers, systems, integrators, and users, research groups, and consulting firms throughout the world that are driving automation forward. I'd like to thank and recognize our partner painted robot painted robot builds and integrates digital solutions.
Speaker 1 00:30:08 They're a web development firm that offers SEO and digital social marketing, and can set up and connect CRM and other ERP tools to unify marketing sales and operations. They can be
[email protected]. If you'd like to get in touch with us at the robot industry podcast, our email address is the robot industry
[email protected]. Or you could find me Jim Beretta on LinkedIn. We'll see you next time. Thanks for listening. Be safe out there. Today's podcast was produced by customer attraction, industrial marketing, and I'd like to thank my nephew, Chris gray for the music, Chris Colvin for audio production, my partner, Janet, and our sponsors <inaudible> and painted robot and new scale robotics.
Speaker 4 00:30:50 <inaudible>.
