The Manufacturing Executive
The Manufacturing Executive

Episode 115 · 1 month ago

Intelligent Robots and the Warehouse of the Future


Our guest today sheds light on what future warehouses will look like as robots become more intelligent.  

Michael Perry is the Vice President of Marketing at Dexterity, Inc. Michael's main objective is to help customers and partners understand how AI-enabled robots can grow their operations. Before Micahel's time with Dexterity, he was Boston Dynamics VP of business development, where he led the go-to-market campaigns for the company's first commercial products. Michael was also the general manager of North America and Director of Strategic Partnerships at DJI, where he built the Enterprise Partnership ecosystem and sales channels for the world's leading drone manufacturer. 

Join us as we discuss:

  • The evolution of robotic intelligence
  • How robots can help solve problems in the supply chain
  • A look into the warehouse of the future 

You have this random flow of huggies and water bottles and captain crunch cereal boxes, and the robot, then, you know, figures out how to stack these on top of each other without crushing the things underneath and creating a stable Palette. It's like playing three D Tetris in a real time, and the robot is just phenomenal at doing it. Welcome to the manufacturing executive podcast, where we explore the strategies and experiences that are driving mid size manufacturers forward. Here you'll discover new insights from passionate manufacturing leaders who have compelling stories to share about their successes and struggles, and you'll learn from B two B sales and marketing experts about how to apply actionable business development strategies inside your business. Let's get into the show, m welcome to another episode of the Manufacturing Executive podcast. I'm Joe Sullivan, your host and a CO founder of the Industrial Marketing Agency guerilla seventy six. This episode is brought to you by Alpha Software Corporation. Alpha software helps manufacturers digitized paper forms, making data collection fast and easy with built in analytics sdashboards. Get a free trial at Alpha software DOT COM. Slash M e. One of my favorite parts about hosting this show is learning about all the incredible technological advancements that are affecting manufacturing today, and nothing blows me away more than the speed at which robotic intelligence is advancing. My guest today is right in the middle of it all, and he's here to tell you about how supply chains will evolve and, specifically, what warehouses of the future will look like as robots continue to get smarter and smarter. Let me introduce him. Michael Perry is the vice president of mark getting at Dexterity, where he helps customers, partners and recruits understand how warehouses can use AI enabled robots to grow their operations. Before Dexterity, Michael was Boston Dynamics Vice President of business development, where he led the go to market campaigns for the company's first commercial products, spot and stretch. He also was the general manager of North America and Director of strategic partnerships at D J I, where he built the enterprise partnership ecosystem and sales channels for the world's leading drone manufacturer. Michael is a graduate of the University of Texas at Austin, with degrees in government and Chinese. After spending years bouncing around Shanghai, Kuala, Lumpur, Hong Kong and L A, he has now settled outside of Boston with his wife and two dogs. Michael, welcome to the show. Thanks, Joe. Great to be here. What's awesome to have you. Well, Michael, we uh, you were talking to me a few days ago when we were kind of prepping for this conversation, about how robots have traditionally worked really well in manufacturing or warehouse settings, where there's consistency among every unit that comes down the line, but when all of a sudden any level of variability is introduced, the robots don't know what to do, and I'm curious to hear from your perspective what's starting to change from an intelligent standpoint to address this. Yeah, that's a good question. You know, many people are familiar with robotic arms. They've seen them in manufacturing contexts. UH, notably if you've been to an auto plant or the past twenty years. These systems are everywhere, but you get a very fixed idea of what these robots are, how they operate and what they can they cannot do. In most contexts, these robot arms are given a very specific task with very confined sets of operations, pick up this item that is the exact same every time and put it into a location that is, you know, microns level, precise in its placement.

Now, over the past few years you've had cobalts or AI enabled robots that give them a little bit more flexibility, but you also see some constraints. They're most notably speed. are very slow. That's because they have to be cautious and working close to people, or they're thinking a lot and trying to say what is the thing that I'm supposed to pick up, how am I going to pick it up, where am I going to put it? In all of these different challenges, we're at a new era of warehouse robotics or robotics in general, where robots have a lot more intelligence to handle the wide variety of things that are presented to them and the wide variety of environments where they can be deployed and be useful. So you're no longer having to say every single time, this is exactly the item that you're going to pick up and you're gonna know its size, weights, dimensions, material conditions every time you touch it and by the way you you're going to put it into this place that is, you know, narrowly constrained. Here you can have all of these different senses, including site sense of touch. Um, you know, some logic to think about how you're going to play something based on what you see in the environment that helps robots react rather than just follow the road set of instructions. All of that starts enabling a wide range of applications that even three years ago, when I first started looking at warehouse robotics, were considered too difficult for automation. Yeah, that's pretty cool to hear. So, as you think about, you know, what the warehouse of the future might look like, Um, what kind of things are you going to see that are are going to be different and changing? Well, there's kind of two thought processes about the way the warehouse of the future would look. One is this idea of the lights out warehouse, where truck pulls up, material gets offloaded, it goes into a large automated storage and retrieval system. Uh, you know, these systems de palletized. They will segment things out, they'll decant boxes into large totes and then eventually pick those totes from those totes into customer orders, they'll be wrapped up and then they get shot out the other side for somebody to put them in the back of a truck. Okay, so that's one way of doing that. It's possible. It's very capex intensive. If one thing goes wrong, then you start wondering about whether or not this system is actually going to have the amount of time that warehouse customers need in order to be successful. The other way of thinking about is is thinking about how to optimize existing human processes with robots. Now there's some downside to that, obviously. With these large s RS is stoms fully automated warehouses, you get a lot of efficiency in throughput, but they're incredibly costly, difficult to maintain and, like I said, the minute there's a hiccup in the system, a person can't go into that process too, you know, start doing the job of of the warehouse itself. We have warehouse robots sit set next to people as they're operating, de Paltizing, simulating, inducting, picking objects into a bend, using the same processes that a person does. You have a lot more flexibility to say we're going to scale up or scale down the number of people are number of robots, depending on the location. If the air hose goes out or the networking goes out in this location, which unfortunately it does from time to time, we still have this redundancy in our operations to stay up. But the challenge with that is it requires a lot more intelligence and flexibility in robots in order to be successful in that...

...environment and we're just now at that point where robots have that full set of capabilities to be successful in those types of human workflows. Let's Micha. Let's talk a little bit about what's been going on in the supply chain and specifically how robots are going to be able to help uh and fix some of the pain that's being felt there over the past few years, or at least how they could. Yeah, absolutely so. One of the big challenges is visibility. Um, you know, I just speaking from personal experience. I'm working on a home renot project and you know these. You know, the windows that we have are delayed by several months, not because of the glass itself, but the hardware components for the glass have been sitting on a dock, we think, in Shanghai for the last four months. So that's what the manufacturer thinks in actually know where where is the good? Is it in Shanghai? Is it on a boat? Is it at the port of Los Angeles waiting to be unloaded? Is that at there? You know their distributor. It's it's really okayue where items are in the current workflow. Now where robots can play an interesting role here is not just alleviating some of the top the first order effects that most people think of in terms of the supply chain of an aging workforce. You have a lot less interest in these physically intense, uh you know, low cognitively rewarding jobs. Uh. So robots can step in and do some of the material handling, these these roles. But the second order effect is really interesting, which is each time a robot picks up something, it can log information about what it's picking up. So it can scan a barcode or you give you material conditions of the package that's picking up. I can tell you how fast it's gone from one side of the warehouse to another just by all the different touches it's had throughout the warehouse. So that data gives you a lot more information about where materials are, how they've flown from one side of the supply chain to another and provide a lot more accuracy in planning. Oh, we know that these items are moving slow slower through the supply chain and we thought this is a good time to start dual sourcing than we then we needed. Or you know what, we're seeing this uptick in demand and we know that these materials flow slower than others, so we're going to order these and bulk before we start getting starved of some of these critical components. So there's this wealth of data that's sort of being collected Um as a result of automation that's just feeding back into the system and helping you to do things even smarter. That's right. Very cool, Michael. Are there any success stories that you can tell us um from, you know, situations where robotics intelligence has transformed the way a company has been operating inside of their supply chain or their facility? Yeah, absolutely. So, you know, we've been a dexterity. We've been working with a number of companies that are looking to apply automation in their Brown field facilities. Again, this is the notion that we've got an existing workflow. We don't have the time, we don't have the CAPEX budget to clear everything out and replace it. You know, all of our operations with automation friendly at workflows. The robots have to fit within the exact the existing manual process and be successful. Now that that's really challenging and daunting, particularly with some of these particular applications that we're looking at. But one of our first big applications is deploying at a...

...bread fulfillment site here in the northeast. So our customer supplies bread to grocery stores across the northeast Um and they needed a robot that was able to fulfill bread orders, picking both individual loaves of bread, tortillas, cookies, chocolate, chocolate cakes, you know, you name it. Had to build these trays with all of these different baked items inside a tray and then also pick, at the same time tray loads of good so you might have one tray that's whole wheat rye, sour dough, hot dump Buns, whatever, and then on top of that is trade, just a hot dog bus and and what they were really afraid of, and looking at the automation space out there, is that for each one of these major skew types they're going to need a separate robot and then on top up to that they're going to need a separate robot to pick at the trade level and then the separate robots to take all of those trades stack them on top of each other, and then a separate robot to roll the trip the conveyor out to somebody to grab them and put them into the back of the truck. This is at the height of the pandemic, when getting people into a warehouse was really challenging. And you know, this company takes a lot of pride and fulfilling their customer demand and I'm sure you remember at that time all of us were fraid, afraid of empty store shelves and so nobody wanted to see bread missing from the store shelves. So that's where dexterity came into the frame. With one robot we were able to pick at the unit level all of this wide mix, I think, over ten thou skews, and each one of those skews is slightly different because bread, as it bakes, it puffs differently and shrink wrap around it morphs in different funny way. Is Flops, the tortillas flopping in a way that uh, cookie box doesn't. So one robot had to be able to pick each one of those goods and pick at the trade level and fit within this existing workflow, which is like a street of all of these different single school goods and all of these quarters that are going back the other side. So that really required a wide variety of cutting edge robotic approaches. Um, you know that that's vision. Um. So being able to look at all this different stuff and figure out what is the edge of this, you know, deform mobile plastic package and, you know, figure out how to grab it. And once it's grabbing it, the robot needs a sense of touch to be able to pick it up without brushing the bread, because I'm sure you can imagine robots love to crush bread, and yet being able to pick something up gently without deforming the deforming the package is important. has to pack it in the tray and then, you know, kind of shuffle the goods together so you're maximizing the tray occupancy, and then figure out how to stack the trays on top of it and to push it out to somebody waiting for it to put into the back of their truck. So that's a success story where you had this very complicated workflow that required a lot. You know, traditionally would require an absurd amount of automation, to the point that it would never make sense financially to deploy all of these robotic systems just to handle this one task. But now we're at a point where one robot can handle the full range of tasks that you would expect in this application and, you know, provide human like throughput rates, uh, to to backfill the workers that couldn't be there during that time or may not want joined because, you know,...

...they're busy doing other stuff. So you know, we're we're tackling other challenges in the space that also required that level of flexibility, whether that's singulating, inducting packages, uh you know, these lava flows of packages that are coming down to shoot, figuring out which one's, Uh uh you know, a carton of curing pods and which one's a deformable poly bag and pushing it onto a trace order. I personally am amazed every time I see our robots, uh Palletie, mixed kew goods, meaning you have this random flow of huggies and water bottles and captain crunch cereal boxes and the robot then you know, figures out how to stack these on top of each other without crushing the things underneath and creating a stable Palette. It's like playing three D Tetris in real time, and the robot is just phenomenal at doing it. So these are the types of complicated, cognitively complex challenges that robots are just now able to start doing, and they're providing a transformative difference for our customers. Let's take a quick break for a word from our sponsor. Still using paper forms for inspections? Alpha Software Corporation helps manufacturers turn paper forms into powerful mobile APPS. You'll create more accurate and thorough manufacturing data, and built in dashboards will help your managers pinpoint quality and supplier issues faster. You don't need to have any development skills to build apps with Alpha software. They offer APP templates that make it easy. Get a free trial at Alpha software dot com, slash m e. that's really cool to hear. So so what you know, as you kind of look ahead, what what's got you most excited about sort of the future of robotics and where things where do you think will be? Things will be had it over the next decade? Yeah, so, you know, I started telling you about how we're tackling some of these tasks individually. And if you break down everything that's in a warehouse, you have, you know, unloading the back of a truck, maybe de palletizing. You have something that's inductating things into storage from storage, and then it's, uh, you know, picking each picking for an order, whether that's at the case level or at the individual goods for e commerce fulfillment, and then it's stacking things on top of each other and then palotizing them and putting them into the back of a truck. Okay, so I've just described all of these different applications individually, but where the magic happens is where you start daisy changing all of these solutions together. So you're bringing the de Palletizing, inducting, the order fulfillment, the palletizing, all with robotics systems, and that's where this data play becomes really powerful. At the beginning of the process, to be into the process, a robot is touching these goods as they're moving through the warehouse. You get a real time snapshot of what's happening in your warehouse facility and as you pop up a level, you can start seeing the real time performance of all of your warehouses across your network and if you're a customer that has a longer supply chain, you can start sharing that data both upstream and downstream to help your partners understand how what is the velocity of goods moving through your system? When can they expect them? You get a lot more transparency and a lot better planning and figuring out how these how goods are moving through your facilities so that you can order correctly. You can maybe cut down the amount of goods that you store at any one time and maybe save some of your warehouse space. where the solver velocity goods that you to have on...

...hand? Tell us a little bit about dexterity. Sarah has been around since twenty nineteen. WE SPUN OUT OF STANFORD'S ROBOT Research Lab and what we do is provide the machine intelligence for robots to handle these complex tasks. We kind of came to the table thinking traditional robot arms is a pretty fixed commodity item. You know, you see a wide variety of robot arms out there. They're phenomenal. O E M is making these robot arms right now, but they're often handstrong in getting scale in some of these applications because they don't have the right vision system, the right since of touched, the right ability to collaborate with other robots on a shared task. So we've created one platform that includes vision, since the touch, collaborative planning, data analytics. What we could do is take all of these individual pieces and script them together so that you can have applications like like, like I was telling you before, the multi skew Palette building, where the robot is looking at this messy pile of goods flowing down a convery belt towards it and it's planning how am I going to pick in place all of these different goods so that I create a stable Palette without crushing things underneath? That requires X, Y Z bit of the Dexterity Platform. Our engineers are able to pull those together into an application. Again, the cool thing about this system it's not just our ability to react to uh large scale applications on the fly Um, but we can also create systems that integrate seamlessly with existing infrastructure and workflows. That means you have very low down time and actually getting out new systems set up. Yeah, we're talking about forty eight hour is to get this multi skew politicizer set up in a new workspace or simulator inductors. So long as you have the right power and calm set up inside your facility, we can deploy in as little as twenty four acts. And Uh, you know, that's a step change from the way that we traditionally think about automation, which requires a lot of work in order to reconfigure spaces for robots to be successful. Using the dexterity platform, using this new intelligence, were able to adapt to our environment and get human like throughput in some of these really complex tasks with minimal site infrastructure changes. It's just amazing to listen to you talk about some of this. It's Uh wild how much has changed and you know, the last five or ten years even, and exciting to think about where things are going absolutely um and the thing that's also really exciting to us is that, you know, it's one thing to do some of these things in Allab and see them tested and validated and you get some really cool results, but where somebody like me gets really excited is seeing them deployed and knowing that, you know, folks here are living on the east coast, if they're going to supermarket and they're buying bread, they're likely buying bread that's been touched by our robots. If you're on the west coast and you're receiving packages from a large e commerce provider and a large retailer, they've likely been touched by our robots and to that's extent we're able to make a transformative difference, not just for our customers but for the people that actually are expecting goods to show up on time and be available to them. And that's a really profound field and that's a more meaningful thing right now than probably it's ever been, just given the way things have played out over the last few years in the world. So pretty cool. Michael's anything us you'd like to add to the conversation that I didn't ask you about... Um, I think the one thing that I it's just a personal passion of mine, is h when, when, when I started thinking about some of the things that really lit up my imagination looking at dexterity for the first time, Um, I started seeing these robot teams working together to accomplish a task. And this is a totally nerdy side tangent, so please bear with me, but Um, one of the things that I thought was really cool was not just seeing these robots individually capable of tackling some of these challenges. Like, again, in our parsonal induction workflow, you just have these packages and boxes and even bowling balls and a back full of earthworms coming down a shoot. In a robot has to figure out how to move things around, to figure out how to grab each one individually and put it on a sorder, tilt trace, sorder. But the real cool thing was starting to see two robot arms doing these tasks next to each other, so they're working on the same shoot. They're able to get the same level of human throughput because, you know, I wouldn't expect somebody to have, you know, the same level of throughput with one arm tied behind their back. So having two arms helps us get at or beyond human parity and throughput. But we start seeing these really novel things where, Um, oversized packages, heavy packages, bowling balls that shouldn't be coming down the shoot eventually do right Um, and you know the level of things that we started encountering that we're like that we're way beyond the SPEC for. You know what we thought a reasonable amount of human intervention should be. So, in a ton aditional sense, you would see one of these heavy packages come down to shoot, the robot would say, I can't pick it up. Somebody come and help me, and then the person would come disentangle this problem. We're saying so much of that that we said, well, the robots need to figure out how to handle this themselves. It's happening at such a high rate that the robots have to figure this out. So what we've done is teach the robots to come together, to pick up the package together and induct it together. So you have these individual agents that are working one after another putting packages on the sorder, but then they come together when they have something that's oversized or too heavy and they'll pick it up, working together to place it on the belt, and then they'll go back to doing their regular workflow. And that's pretty magic, where you have not just the individual intelligence of a robot but then the shared intelligence of the robots working together. Right now that's being, you know, done in a fixed way, right so you have these robots lockdown and they're working from a fixed pedestal. But you can imagine a world where these robot arms are mobile and you know, going down the a line on a the A G V or M R one says, Oh, I need to help with something and they call a Roboto, a robot arm rover, to help them lift something or move something. That's a future that we're moving towards pretty quickly and it lights my imagination on fire with all the possibility. So I can see I can see the passion just from watching you talk here. It's got to be pretty exciting to, you know, be behind the scenes watching all this unfold. Absolutely well, Michael, great conversation today. Can you tell our audience how they can get in touch with you and also where they can learn more about dexterity? Absolutely so, Um, you can reach us at dexterity dot AI. Um We have more information about the exter platform,...

...some of the solutions that we have in singulation, induction, Um Mixqu Pallett, building and so on. Business on Linkedin, we've got, I hate, a tutor on home but we've got some pretty cool videos of robots doing interesting stuff. So swing by the Linkedin page to to see what we're up to. Also, please feel free to reach out to me, Um, Michael Patrick Perry, on Linkedin. I'm always excited to talk about robots Um like, I can't imagine a scenario where somebody's said, hey, I just want to learn more about this industry and what robots are capable of. I always love those conversations and I always learned from as much as I end up sharing, so that's a little bit of the selfish part for me as well. So yeah, please reach out and always easier to chat perfect well, Michael, thanks for doing this today. My pleasure, Q Joe, and as for the rest of you, I hope to catch you on the next episode of the manufacturing executives. Before we go, I want to say a quick thank you to our sponsor, Alpha Software Corporation. Alpha software helps manufacturers digitized paper forms, making data collection fast and easy with built in analytics dashboards. Get a free trial at Alpha software DOT COM. Slash M e. you've been listening to the manufacturing executive podcast. To ensure that you never missed an episode, subscribe to the show in your favorite podcast player. If you'd like to learn more about industrial marketing and sales strategy, you'll find an ever expanding collection of articles, videos, guides and tools specifically for B two B manufacturers at guerrilla SEV dot com slash learn. Thank you so much for listening. UNTIL NEXT TIME.

In-Stream Audio Search


Search across all episodes within this podcast

Episodes (121)