Guests from DHL Supply Chain and Boston Dynamics discuss how the two companies are teaming up to further automate warehousing in North America.
Special Guests:
Guests from DHL Supply Chain and Boston Dynamics discuss how the two companies are teaming up to further automate warehousing in North America.
Special Guests:
Welcome to "All Business. No Boundaries.", a collection of supply chain stories by DHL Supply Chain, the North American leader in contract logistics. I'm your host, Will Heywood. This is a place for in-depth discussions on the supply chain challenges keeping you up at night. We're breaking beyond the boundaries that are limiting your supply chain. Let's dive in. Today's episode is "Stretching the Limits of Supply Chain Digitalization: A Conversation with Boston Dynamics." Joining me are Marc Theermann, Chief Strategy Officer, Boston Dynamics, and Sally Miller, Chief Information Officer, DHL Supply Chain North America. Welcome to you both. Okay. Marc, maybe starting with you, could you give a brief intro of yourself, your role at Boston dynamics, and a little bit about your company?
Marc Theermann:Yeah. Thank you so much, Will. My name is Marc and I'm the Chief Strategy Officer. And for us, that means that I run our products, partnerships, and marketing organizations.
Will Heywood:Okay, great. Sally, same question for you.
Sally Miller:Sure. Sally Miller, I'm the CIO for North America for DHL Supply Chain. I've been with the company for 16 years now. In addition to the traditional IT day-to-day activities, I have responsibility for bringing innovation to the business. So talking with Boston Dynamics today is a great example of what we're doing in the company to address some of the challenges in the supply chain.
Will Heywood:Great. So I guess this the first commercial purchase of stretch in the market, and we're really excited about it, but I'd like to ask you Sally a little bit more about the background on this and how we got to today.
Sally Miller:Sure. The first time we met with Boston dynamics was back in 2018. We were touring some robotics vendors in the Boston area and getting a better understanding of some of the start- up activity going on in the industry and what potential products would be a good fit for our business. And Boston dynamics has long been known as the company . They have some of the industry leading talent in that area, and they were at the time looking to develop products for the supply chain industry. So we instantly had a connection. We were wanting to collaborate with a company and work with someone on getting solutions into our facilities that would improve our operations, excite the associates in the facility, and really paved the way for the future. And Boston dynamics was looking for input in where our pain points were and where the opportunities were. So that's how it really started back in 2018.
Will Heywood:Okay, great. So Marc, I know Boston Dynamics has sort of a wide lens when it looks at different applications for robotics. So I wondered if you could start there and then sort of narrow in on how did you guys start to focus in, on supply chain and where has that brought you?
Marc Theermann:Yeah, I think you will. Boston Dynamics has really been a leader in mobile robotics for many years and the world has seen the funny YouTube videos of our quadruped robots. And we now have three product lines. The first one is our quadruped robot, Spot, which is used for industrial inspections. Then we have a research product line called Atlas, which is a humanoid robot that looks like a human. And then, four years ago, we started to look at where can we combine mobility with manipulation, which we think is the next frontier for robotics. And so we engaged with Sally's team and DHL. And we quickly realized that the form factor that we had in mind at the time was actually going in the wrong direction. And so together with DHL and other companies in the industry, we really honed in on the current design of the Stretch robot, which is a mobile base that can autonomously roam around in a warehouse, but it has a perception mask that can see the warehouse and it has a gripper that can manipulate packages in the warehouse. So we're really combining 30 years of mobility research with now the ability to grasp and manipulate our environment.
Will Heywood:Okay . So , just to press on that a little further; the spot solution looks like, I've heard people describe it as a dog. And I think if you see the latest Katy Perry video , the dog was in there. Spot was in there. Can you describe physically what does Stretch look like?
Marc Theermann:Yeah, you're right. Spot does look a little bit like a dog. And in addition to the Katy Perry video, you have recently maybe seen him in the latest Star Wars episode of Boba Fett, which was near dear to my heart. But stretch, it is really a purpose-built robot. And as such , the anthropomorphism is not particularly strong. Stretch really is a mobile base that is the size of a pallet and on the mobile base, which in essence is an AMR that can autonomously roam around the warehouse. We've built a big arm and the arm is strong enough so that it can handle packages up to about 25 pounds. And it can move those packages at a fairly rapid rate. So we needed a strong arm. So we built something completely custom instead of buying something off the shelf. And we mounted this strong arm on this mobile AMR base. So, envision five pallets stacked on top of one another with a large mobile arm that contains a gripper that has a vacuum suction grip to pick up packages.
Will Heywood:Okay, got it. And Sally, when you and Marc and the Boston Dynamics team were talking, what applications were you thinking were the right purpose for this kind of a robot?
Sally Miller:Any type of activity in the warehouse that is repetitive manual labor intensive is what we were targeting. So it makes the work and opportunities for work for the associates in the warehouse much easier if the heavy lifting is done by an AMR. So our first use case is trailer unloading. So a trailer backs up to the warehouse store and we're able to use Stretch to unload the cartons and put them on a conveyor or stack them or whatever is needed for the process to be completed. It's a fairly manual intensive activity. And the ability to automate it reduces our dependency on labor and improves the work that is done by the other associates in the warehouse. We have found that associates like to work with robots. They like to work in environments that have high tech and it helps us recruit people and retain people.
Will Heywood:So Sally, the trailer and unloading process today, can you describe how that works specifically and how many people do it? What does it look like to get a trailer unloaded?
Sally Miller:Well, it depends on the time commitment needed to get the trailer unloaded, but it is very intensive and can be difficult when the trailer, hopefully, is stacked completely full for associates to get the top layer off and then start working their way through the trailer. So it takes several people and usually a couple hours. And depending on if they're stacking it on pallets or floor-loading it, there are additional steps after the carton is unloaded from the trailer.
Will Heywood:So in the 2B world here with Stretch, do people totally come out of the trailer unloading process? That's a question for both of you, I guess.
Sally Miller:Yes, they do, because we can, as I mentioned, we can stack the cartons onto a pallet or directly onto a conveyor to feed them into automation in the facility.
Will Heywood:So Marc, for all of us who are not roboticists, how do you even start with trying to design a machine that can do something like trailer unloading, which sounds like a fairly complicated physical task to do?
Marc Theermann:Yeah, you're right. It's very difficult. And when we were starting to talk to Sally's team, we noticed that many other companies had tried in the past with multiple different solutions. The way that we approached it is we realized that we wanted to be autonomous, meaning that the robot should completely, autonomously be able to unload the trailer without a human remote controlling the robot. So we needed the right types of sensors to allow the robot to do this. So the robot is equipped with lidar sensors that help the robot, see the environment and understand what is the trailer, what is the warehouse , and what is the conveyor. And then we also noticed that we need to independently see versus grip. So the design that we came up with has a robot that has a perception mask that is independent of the gripper. That way we can quickly move packages back and forth with the gripper. And at the same time independently, continuously monitor the trailer and all the content. So what happens today is that the perception must take continuous photos of all the packages that are in the container and trailer. And then we have machine learning algorithms that can uniquely identify what is a package versus what is not a package. And that is actually a fairly hard robotics challenge. And we were able to combine those three. So the autonomy of the base, the perception mask that can uniquely identify a package, and then a large gripper that is able to grab those packages and put them onto the conveyor belt. And so humans, when you watch them do this today, are actually very good at this task. It just turns out that it's one of the most horrible jobs in the warehouse because it's dark in the summer and hot. It is cold in the winter. There's no windows. And ergonomically, it's often a dangerous environment. So the notion of robots should be deployed where it's dull, dirty, and dangerous, completely holds true for this particular application that a robot actually turns out can be very good at.
Will Heywood:Got it. Got it. That's really interesting. So, Sally, how does the robot know sort of what to do with the cases or the boxes as it unloads them ?
Sally Miller:I don't know the answer to that. Yeah .
Will Heywood:I stumped you .
Sally Miller:I would say that's a better question for Marc, but yeah, you did stump me.
Marc Theermann:Yeah. The thing is , we're starting with this particular application truck unloading because it requires very little integration into the rest of the DHL IT infrastructure, because at the moment the robot does exactly what a truck unloading human would do. It loads, unloads the container onto a conveyor belt, onto a flex conveyor or a fixed line extendable conveyor. And the robot doesn't know what packages it has unloaded. So it doesn't read the barcode or it doesn't know anything about the content of the package itself. So it does not have radar or any sort of x-ray vision built into it yet. So what we are envisioning then, at the moment, the robot unloads the packages onto the conveyor, which is then attached to a scan tunnel, which officially scans the package into the WMS system of the receiving warehouse.
Will Heywood:Okay.
Sally Miller:As we progress to other use cases, there will be an integration between the warehouse management system and Stretch to give it instructions on where to go, how many cartons to pick up and that sort of thing. But right now, for the first use case, it is a nonintegrated application.
Will Heywood:That's where I was going--how does it connect into the WMS? And that comes with the scan at this point. So right now the robot just understand that there are packages in the truck that need to be unloaded and put on a conveyor and that's sort of it.
Marc Theermann:That's right. And the robot picks up a package that he has identified as probably a package. And he picks up the first package and then he takes a look at the package and he says, okay, now I know the full dimensions of the package after I've picked it. And then he assumes that the next package will be the same size and weight so that we can speed up the operations as we go into the deeper into the trailer. But of course, if the package happens to be a different size, then again, the robot looks at it and builds a library of all the packages that might be in the container.
Sally Miller:And most of the time, for our profile of operations, as an example, a large retailer, the inbound cartons are fairly uniform. So by Stretch remembering the typical configuration of the carton, it can move faster. In other applications, across other divisions of DHL, such as Mail or Express, when you have different carton dimensions, it wouldn't move as fast is it would with a single customer dedicated location.
Will Heywood:Okay, understood. So Sally, when you guys are looking at initial rollout within DHL, what does it look like? How many robots would be in a given building? Is there a sweet spot for that? Or do you know yet?
Sally Miller:We will deploy as many Stretch units as are needed based on the inbound volume and the use case. So it could range from one to several. It's all about the profile of that operation. Our plan is to do approximately two or three sites the first half of the year, and learn from that while we're developing other use cases for Stretch. We see tremendous value being able to use Stretch for other things in the building as well as trailer unloading.
Will Heywood:Marc , building off of that. How do you guys see Stretch being applied in the world in the time to come?
Marc Theermann:It's really built as a multipurpose mobile robot. So truck unloading is the first of many applications, but together with our friends at DHL, we certainly have plans to go deeper into the warehouse and expand into different applications as time goes on. So this is not a robot only for truck unloading. This is a robot that starts with truck unloading.
Will Heywood:I want to shift gears a little bit to the partnership between DHL and Boston Dynamics and as a technology provider and a operations provider or solution provider, Sally , you mentioned this started in 2018. Can you guys talk about sort of how these relationships evolve and what you look for in a partner respectively and how these things build up over time to get to where we are now?
Sally Miller:Sure. I can talk specifically to the Boston Dynamics partnership evolution. It started with just a personal connection with the CEO, Robert Playter, and I really liked his openness to understanding the operational part of what we do. At the same time, we were learning how robotics are developed and built and trying to leverage their vast experience in the field. So it started out just as, hey, let's take a look at some of your sites, see what you do, better understand it, and what can we develop to help you? So it evolved from there. And as Boston Dynamics started developing prototypes, we would try them out in our operation. They would watch the performance, take them back into the lab, tweak, develop, and it's been very exciting and gratifying to see the evolution from the first prototype to the actual product of Stretch as it is today. And like I've mentioned, I think there are more use cases for stretch and potential other applications in what we do in supply chains. So we'll continue to have the open dialogue, and as Boston dynamics gains more experience and education on our pain points and can develop solutions, we will collaborate, co-develop, and deploy for our customers. That's the plan.
Will Heywood:Marc, how do you see it? Or how have you been involved?
Marc Theermann:Yeah, it's been an amazing journey with DHL, and I think it shows two things. Number one is that DHL is really such an amazing partner for us because they realized that we needed to go on this journey together and build this robot together and build the specifications for robots like this. And it also shows the deep research commitment that Boston Dynamics has and the heritage that we have in research and the way that we approach building custom robots for these types of applications. So, it's been an exciting couple of years together coming to this point where we now have a robot that is near its final production version. And seeing that deployed at DHL is really a dream come true for many of our engineers that have been working on this project for such a long time.
Will Heywood:So how long does it take to develop something like Stretch from start to finish in order of magnitude investment that you guys put into these technologies?
Marc Theermann:It takes a long time. We're getting better at it, but I would say that there's first a research part that could be, let's say a year long or so . And at the end of that process, we have a deep understanding of the challenge and the problem that we're trying to solve. That's then followed by building a prototype to test out if our hypothesis is actually correct. And if it is correct, and we are marching in the right direction, then maybe it's another year or so before we get to a production version and then maybe another couple of quarters before we could really start rolling out. So often, we're talking a hundred million dollar commitment or so to develop a robot from scratch, but we're getting better at it and we're getting faster at it. We are now thinking of developing a fourth product line at some point. And so we're going through those motions right now.
Will Heywood:Okay. A hundred million. That's a big number, but it makes sense when you kind of lay out the details there. So I get the whole wanting to automate out the dim, dull and dangerous, as I think you put it, Marc. But Sally, from an economic standpoint, there has to be a payoff, right? So from DHL's perspective, how do you evaluate business case return on investment? How does that factor into your go to market approach?
Sally Miller:Well, we consider a lot of things. Obviously, the cost of recruitment of hourly labor retention, simple things like time off all are factored into the labor equation of what we need to do to meet the service level of our customers and compare that versus the cost of the unit and potential use cases to make sure that we do have a use case or a return for the investment.
Will Heywood:Good. Good. Okay . So looking into the future, we've talked a little bit about wanting to get Stretch going on the trailer unloading. I know you're exploring some other use cases, but I think whether you're talking about Stretch in specific or the warehouse environment more broadly, where do you see robotics heading in the future? Sally, why don't you go first?
Sally Miller:Sure. I see robotics more and more as products that can do activities to replace the need for hourly labor during surge periods and to do the type of work that is repetitive and high volume to take that workload off our hourly associates. I do see multiple solutions under one roof. So it'll be interesting to see the orchestration of those solutions to optimize their effectiveness and get the best return out of the investment in the asset. So we'll be looking to WMS vendors and point solutions and the robotics vendors themselves to ensure we're getting the most out of the solution when we have multiple deployed in a building in the future. And I see that in the very near future. I also think that it changes the critical nature and uptime availability of the units, because now in a more manual type operation, you can add more people to get the work out the door, but when you haven't hired them because you have all or a portion of the work being done by robots, it becomes more and more critical that they be available and up. So it's moving us more into a manufacturing like environment, and I'm excited for those new challenges and seeing our sites embrace the technology and get the most out of it. That's a big learning that we've had on the journey is we're able to work with companies like Boston Dynamics and improve their products to make them better so that we get more value out of them. And our role as the operator continues to expand as an integrator and the best selector of solutions based on the profile of the operation that we're running
Will Heywood:Marc, what does the future look like to you?
Marc Theermann:There appears to be two paths to more automation in a warehouse environment. There's the all-in path. When you rip out everything that you have, and you start from scratch with a full ASRS solution that handles everything that happens inside the warehouse, depending on the warehouse that might be, a 50 to 80 million dollar investment. And I do think that that's an exciting path. And whenever you go visit one of these warehouses, it's awe-inspiring the automation that happens there. But we also think there will continuously be a path of folks that don't want all this fixed investment and bolt things to the ground. And I think that's where our robots will play a stronger role because our robots are mobile. And if you don't like them in one location anymore, you can quickly put them in a trailer and drive them somewhere else. In our solution, pretty much nothing gets bolted to the ground. So we think there future where a case handling robot like Stretch, in conjunction with AMRs and autonomous forklifts, could handle almost all the tasks that are needed in a warehouse with almost nothing bolted to the ground. So that's certainly a future that we are marching towards.
Will Heywood:Yeah, that's an exciting one and a long way away from the traditional definition of the supply chain industry, which I think that all three of us came to know as we started our careers a long time ago. Well, thanks to both of you for joining today. I know we're very excited about the partnership that's being announced here this week, and it's the culmination of a lot of hard work done by both DHL and Boston Dynamics. And look forward to hearing about how things go in the future. Hope to have both of you back with a report in a year or two.
Marc Theermann:Thank you, Will. We couldn't be more excited. And thank you, Sally, for the opportunity, not only of course for this partnership, but also to participate in this podcast.
Will Heywood:Excellent. Thank you. If you enjoyed the conversation today, please share it with a friend and rate us on Apple Podcasts. You can find us online at dhl.com/allbusinessnoboundaries and follow us on LinkedIn and Twitter @ DHLSupplyChain. We'll see you next time.