Automating the on-demand economy: keeping up with Amazon’s 200,000 robots

robots warehouses on-demand economy amazon

Digital retail giant Amazon has over 200,000 robots helping deliver more than 350 million different products in an unceasing flood of billions of deliveries. Its fulfillment machine with both free and fast shipping has become a key competitive moat against other retailers: free shipping and 1-day or 2-day shipping is why Amazon customers chose Amazon.

So how can other retailers, whether giants like Walmart or smaller brands, compete? One way is by stealing a march on the e-commerce behemoth and automating themselves.

Support TechFirst: Become a $SMRT stakeholder

In this episode of TechFirst, we chat with Locus Robotics CMO Karen Leavitt. Locus Robotics robots-as-a-service will pick more than a billion items in warehouse, fulfilment centers, and logistics hubs this year alone … and the company just added bigger and more robots. What’s next?

Keep scrolling for full video, audio, and transcript … or check out the Forbes story here.

(Subscribe to my YouTube channel)

TechFirst podcast: robot warehouses to automate the on-demand economy (and compete with Amazon)


Transcript: Locus Robots says it boosts warehouse worker productivity 200-300% via its robots

(This transcript has been lightly edited for length and clarity.)

John Koetsier: When will robots be a real factor in reducing manual labor and freeing humans from boring, repetitive, maybe even dangerous jobs? Well, we might be there right now. Locus Robotics says that its warehouse robots have picked 650 million packages to date and will hit its billionth this year.

Right now we’re joined with Locus CMO, Karen Leavitt. Welcome!

Karen Leavitt: Hey, John, how are you doing? Nice to meet you. 

John Koetsier: Doing really well, thank you for joining us here. Maybe set the stage … what do your robots do?

Karen Leavitt: Okay, well, first I’m going to correct you because we’ve now picked about 705 million. 

John Koetsier: [Laughing] Stop working! You’re going too fast, you’re making me wrong. 

Karen Leavitt: I know, those darn robots. So what do our robots do? The best way to characterize that is to put it in a context everybody understands.

When you order an item online and it gets transmitted to the merchant, in fact what’s happening is your order is getting transmitted to a warehouse. And in that warehouse, the raw material that’s providing you with the merchandise you just ordered … is human labor. So somebody is looking at the order you just placed, going to a specific location in the warehouse where that item is stored, picking it off of a shelf, putting it into sort of a glorified shopping cart, and then shipping it out the door to get to you.

And what our robots do is they help eliminate several of those steps in the human process that makes the job much easier for the worker.

So our robots know what the item is. Nobody has to look at a list. The robots go to the location where the item is being stored and then a worker meets the robot there.

And by doing it that way, we are doubling or even tripling the productivity of the humans in that warehouse and we’re cutting down on the amount of walking that they do by probably 75 or 80%.

These are people who, without the robots, would be walking 10 to 15 miles a day, and now they’re down to just a few miles a day because they’re interacting with the robots. So what do our robots do? They make people’s lives easier and they get you your merchandise faster, and they do it in a predictable way that makes it profitable for their business operators. 

John Koetsier: Now, you said something about your robots, the people meet your robots there — where’s that meeting point? Is that after the robot has gone into the warehouse, done the pick, come back to where it gets packed or shipped or something like that?

Karen Leavitt: Yeah, so the merchandise is stored throughout the warehouse. So imagine hundreds of thousands of square feet of bookshelves — not bookshelves, they’re racks…

John Koetsier: [Laughing] Yes.

Karen Leavitt: … that contain merchandise. And the robot is going to go directly to the location where the item that needs to be picked is stored. The workers, who would ordinarily be serpentining through the aisles pushing a cart, are actually — because we’ve got the robots now — they’re stationed in locations in that warehouse near the racks.

And then when they see a robot stop, they walk over to the robot just a few paces and they follow the instructions on the screen to pick the right item, scan it and drop it into the robot’s container, and then the robot will move to the next pick in the order — we call them ‘picks’ when people make the selection off the shelves.

The robot will go to where the next item needs to be picked off the shelves, and that could be on the other side of the warehouse where it may be met by an entirely different worker. Under ordinary circumstances, before our robots were introduced, the workers would start their journey by loading up a cart with empty containers and then they would walk all through the warehouse, so to get that next pick they would have to walk through the warehouse, but now the robot does that. 

John Koetsier: I have some experience with that because when I was in university, I worked for a while overnight in a warehouse, sometimes picking and shipping stuff, sometimes building stuff, other things like that.

And you have lots of challenges with that, including the traveling salesman problem, of course, right, like what’s the most efficient route to get all of these 25 different items that I need? Humans are not good at that.

At least this human isn’t, and I don’t think humans are generally.

It’s a really challenging situation, so it’s kind of cool if you can get the robots to do that and they’re connected to the operating system essentially of the warehouse to know what to do, where to go, all that stuff. Now you had a bit of a moment last year around Black Friday, Cyber Monday, one of the busiest e-commerce sections of the year … tell me about that. How many total packages did you pick during that week or couple weeks? 

Karen Leavitt: Yup. We picked about 120 million units. Our robots assisted their human coworkers in picking about 120 million units in Cyber Week this year.

Just to put that in context, our robots helped pick about 380 million for the entire year, so you can see that that peak season is such a huge part of the total business for the year for our customers. And that’s up from about 70 million during the same period a year earlier.

But one of the things that was interesting this year was that people started shopping earlier. Instead of all of their orders crammed into that period from Black Friday to Cyber Monday, people started ordering earlier so that they could ensure that their packages would get to their loved ones in time for the holidays. So it was an extended period. So, really you could think of ‘peak’ as having been even bigger than that jump up from the year before. 

John Koetsier: That is kind of fascinating, actually, one in four packages that you picked in 2021 was during that period of Cyber Monday, Black Friday, in and around there. That’s a massive, massive spike that has got to be hugely challenging to staff for, and I guess to “robot” for as well. 

Karen Leavitt: Well, that’s exactly right. So this underscores the value that we provide to our customers. So who are our customers? Our customers are the people who run the warehouses where that merchandise is being stored. And it’s not always consumer goods.

When we talk about Black Friday, of course, we’re talking typically about e-commerce merchandise that people are ordering for consumer purposes. But the same could be true of merchandise that’s… well, not really merchandise… medical devices, pacemakers and joint replacements that are stored in a medical device warehouse, or auto parts in a parts warehouse. But all of these things have one thing in common: they’re stored in one big, large facility and they have to get out to their final consumers — whether those are business consumers, medical consumers, or personal consumers — in a timely fashion.

So the folks who run these warehouses, they’re really experts at leveraging the supply chain. This is the heart of the supply chain economy.

And in most of those warehouses, 95% of all of those warehouses do this process entirely manually, where it’s a person pushing a glorified shopping cart through the aisles, walking, as I said, a dozen or more miles a day, to get this merchandise.

And demand for e-commerce picks or online-ordered picks has increased dramatically over the last decade, but of course during the pandemic it spiked up even more sharply. Labor has become more difficult to find and frankly, it’s a really challenging job. It’s exhausting. So what our robots do is they help make that job not only easier for the workers by reducing their walking time, reducing the number of errors involved, just making it easier so they’re not pushing heavy carts … but it also delivers a lot of actionable intelligence to the business operator.

We are giving our customers an enormous amount of information, not just reams of spreadsheets, but dashboards and things that they can actually take action against right this minute. So they can know where you direct the workers in the warehouse. They know what their labor requirements are going to be. And then because the robots operate as a coordinated fleet, when our customers’ business reaches this peak level, we ship them more robots. They hire more temporary workers. At the same time, they hire more robotic workers.

And so we will ship them additional robots for just a two or three month period, during the peak season, and then in January they ship those extra robots back to us. So our customers are really getting just-in-time and just-on-budget resources to get done the job that they need to, to serve their customers.

John Koetsier: Well, that’s pretty interesting, because of course when you bring on temporary labor, you’ve got to train temporary labor. Humans don’t instantly know, ‘Oh, that’s where everything is. That’s how I do it, that’s how you do things here.’ The robots, on the other hand, plug them in and there you go, right? 

Karen Leavitt: That’s exactly right. It takes about three minutes — and that’s probably generous — from the time that the carton is opened on the loading dock and the robot joins the rest of its robot friends to start working. The robot immediately latches onto the robot network and starts to work, and then our software platform knits together all of the instructions. We interface with other systems in the warehouse to take orders that are coming in from consumers at home.

And we give direction, not just to the robots about what to do next… ‘Hey, robot, step forward, we’re going to send you out on a mission to go get some merchandise,’ but we also instruct the workers on what they’re supposed to do. So you raised a really great point that typically in an un-automated warehouse, it takes about two weeks to train a new worker, whether the worker is permanent or temporary.

But with the robots, the robots are not just carrying merchandise through the warehouse, they’re also carrying instructions for the workers. So the training for a worker is, ‘Hey, you see that robot over there? When you see it…

John Koetsier: Do what it tells you to do [laughter & crosstalk].

Karen Leavitt: … walk over and do whatever it tells you.

John Koetsier: Your robotic overlords are now in charge. 

Karen Leavitt: Exactly. Do whatever it tells you… well, they’re friendly. They’re very friendly, and they use lots of pictures and a very familiar interface, so it really makes people’s jobs easier. And we have not just the executives who are making the buying decisions, but the workers in the warehouse themselves tell us all the time how much they love working with the robots.

It’s really a sight to behold when you see hundreds of robots and dozens of workers working side by side in a warehouse.

John Koetsier: So let’s get into that in just a bit, what that looks like and how that will evolve over time. Before that, what gets better when you add robots into a warehouse? How much faster are you picking things? How much more accurate are you? Are there fewer injuries? 

Karen Leavitt: All of those things. So, we suffer from sort of having a laundry list of benefits, which is, I guess, a really good problem to have

John Koetsier: [Laughing] It’s a horrible thing to suffer from.

Karen Leavitt: Yeah. Our customers are constantly quoting this litany of benefits. So, starting right out of the gate with productivity, we typically will double or triple the rate at which the workers can do their job. We do that of course by… and in addition to making the worker’s job easier, we’re also making them much more productive.

And we can do it very predictably so that our customers can plan ahead for what their capacity is going to be, and really be able to count on that. As far as error rates go, well, it’s the robot that’s checking to make sure that you’ve picked the right item off the shelf. Warehouses can be crowded places…

John Koetsier: Or that the robot has picked the right item off the shelf? 

Karen Leavitt: That the… I’m sorry, that the worker has picked the right item off the shelf, but the robot is verifying that. So warehouses… 

John Koetsier: So I thought the robot was picking the item off the shelf?

Karen Leavitt: The robot doesn’t do the picking. The robots don’t have arms. I actually… 

John Koetsier: Oh, okay.

Karen Leavitt: I have a robot right here. I have a miniature of the robot right here [holds up mini robot], but the robot moves around on wheels through the warehouse using a series of cameras and laser beams. 

John Koetsier: So then you have to have workers stationed at various points and the robot basically sends up the bat signal, it starts flashing or something ‘help me out’ and it says, ‘Hey, pick that thing, put it on my deck’ type of thing? 

Karen Leavitt: That’s exactly it. So the robot will go to the merchandise location. It will sit positioned just next to where the item is, so any worker glancing at it knows immediately where the item to be picked can be found. And then the robot has on it an iPad, a touch screen which will display a picture of the item and say the exact location.

The worker will pick from that location because people are still best at this mov— [grasping hand gesture], you know, this motion. The worker will pick it, we’ll scan it on the scanner that’s built into the robot, and we’ll put it into the robot’s container. Could be a whole variety of different container types, and…

John Koetsier: And then it’ll take it to a different station where somebody else will put it…

Karen Leavitt: That’s exactly.

John Koetsier: … take it out and put it on a box or something like that. 

Karen Leavitt: That’s it.

John Koetsier: So you’re going from having a scenario where you have people wandering around picking, then taking somewhere, then packaging, then shipping, to a scenario where you’ve got some people spread around the warehouse looking for robots in need of help, putting the thing on the robot, the robot goes off to shipping and receiving, it gets packaged, boom, and out the door. Is that correct? 

Karen Leavitt: That’s correct. And most important, remember that for the warehouse, the warehouse generates most of its revenue from the act of that pick, when it’s taking something to ship to a consumer is when the top-line revenue comes in.

So what we’re doing is we’re taking workers and putting them on the most revenue-intensive part of the job, which is the pick, instead of the time-consuming, wasted, non-revenue part of the job, which is walking through the warehouse. 

John Koetsier: That is interesting though, because that’s going to be a change, right? That’s a significant change, because if you’re running a warehouse and what you’re probably going to do is you’re going to have… I don’t know what the number is, but you’re going to have some people who are hanging out in your aisles looking for those robots who need a bit of assistance, and then you’re going to have some people elsewhere. The time you’re saving is the person pushing the cart around and walking the 15 kilometers or 15 miles a day or whatever the case might be. But that is a bit of an interesting change isn’t it?

Karen Leavitt: Well, it isn’t and it isn’t. So it’s not as if there are hoards of workers hanging out on the aisle street corners waiting for a robot to come along. 

John Koetsier: One for every bay [laughing].

Karen Leavitt: And actually it’s usually not every bay. So this is where the intelligence comes into the process, that we’re able to work together with the other systems in the warehouse to predict when the loads of orders are going to come down, what the workload is going to be required of both robots and workers, so our customers will staff appropriately. If it’s the type of business where a lot of orders come in overnight, then they’re going to be fully staffed early in the morning. If they tend to get orders early in the morning, in the past, they may have had to bring workers in early just to clear whatever backlog so they’d be prepared for the crush.

But now they can have their workers report to work just in time and be there for when the large orders come through. But then our software does something magical. We don’t just send orders out all through the warehouse.

We don’t send robots necessarily all through the warehouse. We selectively choose the orders that we’re going to be picking right this second, and cluster orders together so that on a robot like this that may have — obviously, this is a miniature version of the robot, but it might have multiple containers on it with different compartments.

So this robot might be picking six different orders simultaneously of, say, supplies to go to a family that’s in the back-to-school area. And we’re going to choose six orders whose contents are located in close proximity to one another in the warehouse. 

John Koetsier: Sure.

Karen Leavitt: So what we’re doing is we’re optimizing for the density of work — density on the robot and density within the warehouse. So… 

John Koetsier: Cool. So let’s talk about the humans then, because I’m firmly a believer that we need more robotics, we need more automation. We want to not have people working in jobs that could be really boring for some — maybe not for all — that tend to wear them out. I mean, we’ve seen the horror stories from workers in Amazon warehouses, for instance. I mean, repetitive stress injuries, back injuries, all kinds of things, just general not doing well because being pushed to go faster, faster, faster, faster and keep up with this incredible pace of stuff that we’re buying stuff online and on mobile.

So I’m a big believer in that, but there’s also the reality that people need to feed their families. People need to pay the rent. And there’s a fear that adding robots means taking jobs. Talk about that tension and how you see that. And if there is a tension, how we resolve it. 

Karen Leavitt: Sure. So I’ll tell you … three, four years ago, I worried about that tension … but we haven’t seen it. And the reason we haven’t seen it is because the growth rate in the fulfillment warehouse area has just been so strong that the ability to find, hire and retain labor is still the biggest challenge warehouse operators face. They’re not operating on the belief that, oh my gosh, I want to lay off my entire workforce.

On the contrary, they’re operating on the principle that I really want to be able to attract and retain the best workers here. And in order to do that, I want to be able to make sure that their jobs are more meaningful. And by pairing them with a robot, the human and the robot get to share the work in a way that’s appropriate for both of them. The human is going to do the thinking and the manual dexterity tasks. The robot’s going to be doing the carrying and the traveling tasks.

And they work together beautifully. If you were to go and visit our website you would see, literally we have a love story video of workers just standing there talking about how much they love working with the robots, how much it’s made their jobs easier. It doesn’t just make their job easier. It means that they can be a better worker in a way that’s more fulfilling for them and more beneficial to the employer, because now, instead of spending their time traipsing through aisles, they can become an expert on their area of the warehouse. They can start to make contributions back, recommendations about best practices for how to optimize their section of the warehouse. 

And what we do is we’re really turning these warehouses into digital command centers. We put up monitors everywhere that create dashboards and we see not just the supervisors and the executives looking at these dashboards, but the workers looking at them. They can see how their actions are contributing to the output in the warehouse, and they can take action as a result of that.

So we’re really enabling that. We also, just to create a little fun, we also build gamification into our robots so that workers can sort of compete either against themselves or against their coworkers to make things a little bit more interesting. They can set targets and see how fast they’re picking. So there’s really a tremendous feedback loop there. And then you’d mentioned earlier, things like reduction in injury rates.

We have one customer who quotes to us an 80% reduction in injuries.

When workers are pushing carts, those carts over the course of a roughly one hour or 90-minute mission of snaking through the warehouse can get up to 200 or 300 pounds. We have a customer in the U.K. who refers to the carts as ‘widowmakers.’ 

John Koetsier: Wow.

Karen Leavitt: And not so tongue-in-cheek. So they see an 80% reduction in injuries, but also you’ve got workers who are just less fatigued. So there’s really a sense of empowerment that comes from this, rather than a sense of displacement.

John Koetsier: That’s interesting to hear, because I recently interviewed the CEO of German Bionic, which makes an exoskeleton, essentially, for manual laborers that takes some of the weight if you’re lifting something, if you’re walking long distances. And some of the feedback that they got was that workers who were dead tired at the end of the day and, for instance, couldn’t play with their kids or do some other tasks that they want to do in their personal life, now have that energy.

So that makes sense.

I hope that the other things also stay true when the robot gets good enough to pick itself from the cart. And I hope that we find ways to integrate humans into that to add value and to do better things as well. Maybe let’s end here … we’re getting better all the time at electronics, at automation, at robotics … what does your solution look like in five years? 

Karen Leavitt: In five years? Well, first, we’re expanding our product family. We will be announcing some new additions to the family in a few months where we’ve got different form factors so that we can carry not just a hundred pounder or fewer items, but hundreds of pounds or thousands of pounds even of merchandise. 

John Koetsier: Wow.

Karen Leavitt: So that we can carry larger objects through the warehouse. We already are starting to expand into more use cases in the warehouse itself, not just outbound fulfillment that is picking, but also returns. Over the holidays this year, people ordered a lot of things that they were going to return, and that trend is continuing … getting things from one dock door to another.

And then of course, yes, there are people all the time who are searching for ways of alleviating the need for human labor in certain, as you said, repetitive and tedious tasks. So the technology does get better all the time and we’re looking forward to the day when that happens.

But the humans won’t disappear. The humans will move on to other tasks in the warehouse. Warehouses are complex digital environments and humans will move into other areas where now that the robots are improving the state of material movement in one area, it means that there’ll be really high demand for jobs of people who are looking to interpret those actions downstream from there. So people will still have roles. 

John Koetsier: Very good. I remember one of the roles that I had sometimes when we were taking a break is we’d go for a little ride on some of the machines [laughing]. I wonder if that happens on the robots? I won’t ask that question, however. But thank you for this time, Karen. I do appreciate it.

Karen Leavitt: Thanks, John. Very nice to meet you.

TechFirst is about smart matter … drones, AI, robots, and other cutting-edge tech

Made it all the way down here? Wow!

The TechFirst with John Koetsier podcast is about tech that is changing the world, including wearable tech, and innovators who are shaping the future. Guests include former Apple CEO John Scully. The head of Facebook gaming. Amazon’s head of robotics. GitHub’s CTO. Twitter’s chief information security officer, and much more. Scientists inventing smart contact lenses. Startup entrepreneurs. Google executives. Former Microsoft CTO Nathan Myhrvold. And much, much more.

Consider supporting TechFirst by becoming a $SMRT stakeholder, connect to my YouTube channel, and subscribe on your podcast platform of choice: