0:00:00.000,0:00:03.561
I want to talk to you about,

0:00:03.561,0:00:07.122
or share with you, a[br]breakthrough new approach

0:00:07.122,0:00:10.464
for managing items of[br]inventory inside of a warehouse.

0:00:10.464,0:00:13.022
We're talking about a pick,[br]pack and ship setting here.

0:00:13.022,0:00:15.845
So as a hint,

0:00:15.845,0:00:19.998
this solution involves[br]hundreds of mobile robots,

0:00:19.998,0:00:22.422
sometimes thousands[br]of mobile robots,

0:00:22.422,0:00:25.020
moving around a warehouse.[br]And I'll get to the solution.

0:00:25.020,0:00:26.240
But for a moment, just think

0:00:26.240,0:00:28.766
about the last time that[br]you ordered something online.

0:00:28.766,0:00:30.830
You were sitting[br]on your couch

0:00:30.830,0:00:34.890
and you decided that you[br]absolutely had to have this red t-shirt.

0:00:34.890,0:00:37.186
So — click! — you put it[br]into your shopping cart.

0:00:37.186,0:00:39.084
And then you decided[br]that green pair of pants

0:00:39.084,0:00:40.982
looks pretty good too — click!

0:00:40.982,0:00:42.882
And maybe a blue[br]pair of shoes — click!

0:00:42.882,0:00:44.932
So at this point you've[br]assembled your order.

0:00:44.932,0:00:46.700
You didn't stop to think [br]for a moment that

0:00:46.700,0:00:48.468
that might not be a great outfit.

0:00:48.468,0:00:50.238
But you hit[br]"submit order."

0:00:50.238,0:00:54.412
And two days later, this package[br]shows up on your doorstep.

0:00:54.412,0:00:57.366
And you open the box and you're[br]like, wow, there's my goo.

0:00:57.366,0:01:00.327
Did you ever stop to think about[br]how those items of inventory

0:01:00.327,0:01:04.540
actually found their way inside[br]that box in the warehouse?

0:01:04.540,0:01:08.358
So I'm here to tell you [br]it's that guy right there.

0:01:08.358,0:01:11.978
So deep in the[br]middle of that picture,

0:01:11.978,0:01:14.493
you see a classic[br]pick-pack worker

0:01:14.493,0:01:17.718
in a distribution or[br]order fulfillments setting.

0:01:17.718,0:01:21.600
Classically these pick workers will [br]spend 60 or 70 percent of their day

0:01:21.600,0:01:23.512
wandering around[br]the warehouse.

0:01:23.512,0:01:26.223
They'll often walk[br]as much as 5 or 10 miles

0:01:26.223,0:01:28.884
in pursuit of[br]those items of inventory.

0:01:28.884,0:01:33.305
Not only is this an[br]unproductive way to fill orders,

0:01:33.305,0:01:37.362
it also turns out to be an[br]unfulfilling way to fill orders.

0:01:37.362,0:01:40.993
So let me tell you where I[br]first bumped into this problem.

0:01:40.993,0:01:45.047
I was out in the Bay area [br]in '99, 2000, the dot com boom.

0:01:45.047,0:01:49.109
I worked for a fabulously[br]spectacular flame-out called Webvan.

0:01:49.109,0:01:50.930
(Laughter)

0:01:50.930,0:01:53.660
This company raised hundreds of[br]millions of dollars with the notion that

0:01:53.660,0:01:56.442
we will deliver[br]grocery orders online.

0:01:56.442,0:02:00.615
And it really came down to the fact[br]that we couldn't do it cost effectively.

0:02:00.615,0:02:04.577
Turns out e-commerce was something[br]that was very hard and very costly.

0:02:04.577,0:02:08.828
In this particular instance we were trying[br]to assemble 30 items of inventory

0:02:08.828,0:02:12.800
into a few totes, onto a van[br]to deliver to the home.

0:02:12.800,0:02:16.600
And when you think about it,[br]it was costing us 30 dollars.

0:02:16.600,0:02:19.686
Imagine, we had an[br]89¢ can of soup

0:02:19.686,0:02:23.362
that was costing us one dollar to[br]pick and pack into that tote.

0:02:23.362,0:02:26.570
And that's before we actually[br]tried to deliver it to the home.

0:02:26.570,0:02:29.357
So long story short,[br]during my one year at Webvan,

0:02:29.357,0:02:32.593
what I realized by talking to [br]all the material-handling providers

0:02:32.593,0:02:37.231
was that there was no solution designed[br]specifically to solve each base picking.

0:02:37.231,0:02:41.254
Red item, green, blue, getting[br]those three things in a box.

0:02:41.254,0:02:44.239
So we said, there's just[br]got to be a better way to do this.

0:02:44.239,0:02:46.766
Existing material handling[br]was set up to pump

0:02:46.766,0:02:50.533
pallets and cases of[br]goo to retail stores.

0:02:50.533,0:02:54.330
Of course Webvan went out of business,[br]and about a year and a half later,

0:02:54.330,0:02:57.910
I was still noodling on this problem.[br]It was still nagging at me.

0:02:57.910,0:02:59.778
And I started[br]thinking about it again.

0:02:59.778,0:03:04.905
And I said, let me just focus briefly[br]on what I wanted as a pick worker,

0:03:04.905,0:03:07.236
or my vision for[br]how it should work.

0:03:07.236,0:03:08.650
(Laughter)

0:03:08.650,0:03:10.572
I said, let's focus[br]on the problem.

0:03:10.572,0:03:14.427
I have an order here and what[br]I want to do is I want to put

0:03:14.427,0:03:16.647
red, green and blue[br]in this box right here.

0:03:16.647,0:03:19.499
What I need is a system where I put out[br]my hand and — poof! —

0:03:19.499,0:03:22.022
the product shows up [br]and I pack it into the order,

0:03:22.022,0:03:23.715
and now we're thinking,

0:03:23.715,0:03:27.729
this would be a very operator-centric[br]approach to solving the problem.

0:03:27.729,0:03:31.956
This is what I need. What technology[br]is available to solve this problem?

0:03:31.956,0:03:35.897
But as you can see, orders can come[br]and go, products can come and go.

0:03:35.897,0:03:40.580
It allows us to focus on making the[br]pick worker the center of the problem,

0:03:40.580,0:03:45.204
and providing them the tools to make[br]them as productive as possible.

0:03:45.204,0:03:47.304
So how did I[br]arrive at this notion?

0:03:47.304,0:03:50.837
Well, actually it came from[br]a brainstorming exercise,

0:03:50.837,0:03:53.677
probably a technique[br]that many of you use,

0:03:53.677,0:03:55.833
It's this notion of[br]testing your ideas.

0:03:55.833,0:03:57.563
Take a blank sheet, of course,

0:03:57.563,0:04:01.593
but then test your ideas[br]at the limits — infinity, zero.

0:04:01.593,0:04:04.775
In this particular case, we[br]challenged ourselves with the idea:

0:04:04.775,0:04:07.751
What if we had to build a[br]distribution center in China,

0:04:07.751,0:04:10.365
where it's a very,[br]very low-cost market?

0:04:10.365,0:04:13.578
And say, labor is cheap,[br]land is cheap.

0:04:13.578,0:04:15.224
And we said specifically,

0:04:15.224,0:04:18.230
"What if it was zero dollars[br]an hour for direct labor

0:04:18.230,0:04:20.946
and we could build a million-[br]square-foot distribution center?"

0:04:20.946,0:04:23.021
So naturally that[br]led to ideas that said,

0:04:23.021,0:04:24.829
"Let's put lots of people[br]in the warehouse."

0:04:24.829,0:04:27.374
And I said, "Hold on,[br]zero dollars per hour,

0:04:27.374,0:04:30.231
what I would do is 'hire'

0:04:30.231,0:04:34.358
10,000 workers to come to the[br]warehouse every morning at 8 a.m.,

0:04:34.358,0:04:37.477
walk into the warehouse and[br]pick up one item of inventory

0:04:37.477,0:04:39.210
and then just stand there.

0:04:39.210,0:04:41.749
So you hold Captain Crunch,[br]you hold the Mountain Dew,

0:04:41.749,0:04:43.172
you hold the Diet Coke.

0:04:43.172,0:04:45.428
If I need it, I'll call you,[br]otherwise just stand there.

0:04:45.428,0:04:48.815
But when I need Diet Coke and I call it,[br]you guys talk amongst yourselves.

0:04:48.815,0:04:52.936
Diet Coke walks up to the front —[br]pick it, put it in the tote, away it goes."

0:04:52.936,0:04:57.588
Wow, what if the products[br]could walk and talk on their own?

0:04:57.588,0:04:59.926
That's a very interesting,[br]very powerful way

0:04:59.926,0:05:02.520
that we could potentially[br]organize this warehouse.

0:05:02.520,0:05:04.868
So of course,[br]labor isn't free,

0:05:04.868,0:05:08.176
on that practical[br]versus awesome spectrum.

0:05:08.176,0:05:09.866
(Laughter)

0:05:09.866,0:05:12.983
So we said mobile shelving —[br]We'll put them on mobile shelving.

0:05:12.983,0:05:17.776
We'll use mobile robots and[br]we'll move the inventory around.

0:05:17.776,0:05:22.194
And so we got underway on that and[br]then I'm sitting on my couch in 2008.

0:05:22.194,0:05:26.013
Did any of you see the Beijing[br]Olympics, the opening ceremonies?

0:05:26.013,0:05:29.054
I about fell out of my[br]couch when I saw this.

0:05:29.054,0:05:30.527
I'm like, that was the idea!

0:05:30.527,0:05:35.026
(Laughter and Applause)

0:05:35.026,0:05:38.917
We'll put thousands of people on[br]the warehouse floor, the stadium floor.

0:05:38.917,0:05:42.952
But interestingly enough, this[br]actually relates to the idea

0:05:42.952,0:05:48.364
in that these guys were creating some[br]incredibly powerful, impressive digital art,

0:05:48.364,0:05:50.440
all without computers,[br]I'm told,

0:05:50.440,0:05:52.886
it was all peer-to-peer[br]coordination and communication.

0:05:52.886,0:05:54.462
You stand up,[br]I'll squat down.

0:05:54.462,0:05:56.107
And they made[br]some fabulous art.

0:05:56.107,0:05:58.795
It speaks to the[br]power of emergence

0:05:58.795,0:06:02.763
in systems when you let things[br]start to talk with each other.

0:06:02.763,0:06:06.541
So that was a little[br]bit of the journey.

0:06:06.541,0:06:10.579
So of course, now what became[br]the practical reality of this idea?

0:06:10.579,0:06:12.385
Here is a warehouse.

0:06:12.385,0:06:16.161
It's a pick, pack and ship center[br]that has about 10,000 different SKUs.

0:06:16.161,0:06:20.298
We'll call them red pens,[br]green pens, yellow Post-It Notes.

0:06:20.298,0:06:23.677
We send the little orange robots[br]out to pick up the blue shelving pods.

0:06:23.677,0:06:25.845
And we deliver them[br]to the side of the building.

0:06:25.845,0:06:28.981
So all the pick workers now[br]get to stay on the perimeter.

0:06:28.981,0:06:31.323
And the game here is[br]to pick up the shelves,

0:06:31.323,0:06:34.647
take them down the highway and[br]deliver them straight to the pick worker.

0:06:34.647,0:06:36.784
This pick worker's life[br]is completely different.

0:06:36.784,0:06:40.201
Rather than wandering around[br]the warehouse, she gets to stay still

0:06:40.201,0:06:41.789
in a pick station like this

0:06:41.789,0:06:45.787
and every product in the[br]building can now come to her.

0:06:45.787,0:06:49.065
So the process[br]is very productive.

0:06:49.065,0:06:53.260
Reach in, pick an item,[br]scan the bar code, pack it out.

0:06:53.260,0:06:54.909
By the time[br]you turn around,

0:06:54.909,0:06:57.948
there's another product there[br]ready to be picked and packed.

0:06:57.948,0:07:00.958
So what we've done is take[br]out all of the non-value added

0:07:00.958,0:07:03.642
walking, searching,[br]wasting, waited time,

0:07:03.642,0:07:07.920
and we've developed a very [br]high-fidelity way to pick these orders,

0:07:07.920,0:07:12.084
where you point at it with[br]a laser, scan the UPC barcode,

0:07:12.084,0:07:15.278
and then indicate with a light[br]which box it needs to go into.

0:07:15.278,0:07:18.764
So more productive, more[br]accurate and, it turns out,

0:07:18.764,0:07:23.070
it's a more interesting office[br]environment for these pick workers.

0:07:23.070,0:07:25.507
They actually complete[br]the whole order.

0:07:25.507,0:07:28.034
So they do red, green and blue,[br]not just a part of the order.

0:07:28.034,0:07:31.648
And they feel a little bit more[br]in control of their environment.

0:07:31.648,0:07:34.846
So the side effects[br]of this approach

0:07:34.846,0:07:36.414
are what really surprised us.

0:07:36.414,0:07:37.992
We knew it was going [br]to be more productive.

0:07:37.992,0:07:42.062
But we didn't realize just how[br]pervasive this way of thinking

0:07:42.062,0:07:47.292
extended to other[br]functions in the warehouse.

0:07:47.292,0:07:52.124
But what effectively this approach[br]is doing inside of the DC

0:07:52.124,0:07:56.588
is turning it into a massively[br]parallel processing engine.

0:07:56.588,0:07:59.211
So this is again a[br]cross-fertilization of ideas.

0:07:59.211,0:08:01.150
Here's a warehouse[br]and we're thinking about

0:08:01.150,0:08:04.669
parallel processing[br]supercomputer architectures.

0:08:04.669,0:08:07.066
The notion here is that you have

0:08:07.066,0:08:09.863
10 workers on [br]the right side of the screen

0:08:09.863,0:08:14.111
that are now all independent[br]autonomous pick workers.

0:08:14.111,0:08:18.015
If the worker in station three decides[br]to leave and go to the bathroom,

0:08:18.015,0:08:21.499
it has no impact on the[br]productivity of the other nine workers.

0:08:21.499,0:08:25.869
Contrast that, for a moment, with the[br]traditional method of using a conveyor.

0:08:25.869,0:08:27.928
When one person[br]passes the order to you,

0:08:27.928,0:08:30.427
you put something in[br]and pass it downstream.

0:08:30.427,0:08:33.544
Everyone has to be in place[br]for that serial process to work.

0:08:33.544,0:08:36.721
This becomes a more robust[br]way to think about the warehouse.

0:08:36.721,0:08:41.265
And then underneath the hoods gets[br]interesting in that we're tracking

0:08:41.265,0:08:43.003
the popularity[br]of the products.

0:08:43.003,0:08:45.827
And we're using dynamic[br]and adaptive algorithms

0:08:45.827,0:08:50.281
to tune the floor[br]of the warehouse.

0:08:50.281,0:08:55.165
So what you see here potentially[br]the week leading up to Valentine's Day.

0:08:55.165,0:08:59.008
All that pink chalky candy has[br]moved to the front of the building

0:08:59.008,0:09:02.901
and is now being picked into a[br]lot of orders in those pick stations.

0:09:02.901,0:09:07.064
Come in two days after Valentine's Day,[br]and that candy, the leftover candy,

0:09:07.064,0:09:09.337
has all drifted to the[br]back of the warehouse

0:09:09.337,0:09:13.630
and is occupying the cooler[br]zone on the thermal map there.

0:09:13.630,0:09:17.113
One other side effect of this[br]approach using the parallel processing

0:09:17.113,0:09:20.108
is these things can[br]scale to ginormous.

0:09:20.108,0:09:21.635
(Laughter)

0:09:21.635,0:09:24.369
So whether you're doing[br]two pick stations, 20 pick stations,

0:09:24.369,0:09:27.642
or 200 pick stations, the[br]path planning algorithms

0:09:27.642,0:09:30.185
and all of the inventory[br]algorithms just work.

0:09:30.185,0:09:34.541
In this example you[br]see that the inventory

0:09:34.541,0:09:36.867
has now occupied all the[br]perimeter of the building

0:09:36.867,0:09:39.213
because that's where[br]the pick stations were.

0:09:39.213,0:09:41.231
They sorted it[br]out for themselves.

0:09:41.231,0:09:43.518
So I'll conclude with[br]just one final video

0:09:43.518,0:09:46.502
that shows how[br]this comes to bear

0:09:46.502,0:09:50.096
on the pick worker's actual[br]day in the life of.

0:09:50.096,0:09:54.432
So as we mentioned, the process is[br]to move inventory along the highway

0:09:54.432,0:09:57.084
and then find your way[br]into these pick stations.

0:09:57.084,0:09:59.555
And our software in the background

0:09:59.555,0:10:02.306
understands what's going on [br]in each station,

0:10:02.306,0:10:04.917
we direct the pods[br]across the highway

0:10:04.917,0:10:07.631
and we're attempting to[br]get into a queuing system

0:10:07.631,0:10:10.515
to present the work[br]to the pick worker.

0:10:10.515,0:10:13.760
What's interesting is we can even[br]adapt the speed of the pick workers.

0:10:13.760,0:10:17.686
The faster pickers get more pods[br]and the slower pickers get few.

0:10:17.686,0:10:20.824
But this pick worker now is[br]literally having that experience

0:10:20.824,0:10:22.677
that we described before.

0:10:22.677,0:10:25.221
She puts out her hand.[br]The product jumps into it.

0:10:25.221,0:10:27.426
Or she has to reach in and get it.

0:10:27.426,0:10:29.871
She scans it and [br]she puts it in the bucket.

0:10:29.871,0:10:33.667
And all of the rest of the technology[br]is kind of behind the scenes.

0:10:33.667,0:10:37.528
So she gets to now focus on the[br]picking and packing portion of her job.

0:10:37.528,0:10:40.802
Never has any idle time,[br]never has to leave her mat.

0:10:40.802,0:10:44.942
And actually we think[br]not only a more productive

0:10:44.942,0:10:48.212
and more accurate[br]way to fill orders.

0:10:48.212,0:10:51.622
We think it's a more[br]fulfilling way to fill orders.

0:10:51.622,0:10:54.830
The reason we can say[br]that, though, is that workers

0:10:54.830,0:10:56.972
in a lot of these[br]buildings now compete

0:10:56.972,0:11:00.154
for the privilege of working[br]in the Kiva zone that day.

0:11:00.154,0:11:02.817
And sometimes we'll catch[br]them on testimonial videos

0:11:02.817,0:11:04.934
saying such things as,

0:11:04.934,0:11:09.180
they have more energy after the[br]day to play with their grandchildren,

0:11:09.180,0:11:13.776
or in one case a guy said, "the[br]Kiva zone is so stress-free

0:11:13.776,0:11:16.891
that I've actually stopped taking[br]my blood pressure medication."

0:11:16.891,0:11:18.725
(Laughter)

0:11:18.725,0:11:22.724
That was at a pharmaceutical distributor,[br]so they told us not to use that video.

0:11:22.724,0:11:26.292
(Laughter)

0:11:26.292,0:11:29.316
So what I wanted to leave you[br]with today is the notion that

0:11:29.316,0:11:32.171
when you let things start[br]to think and walk

0:11:32.171,0:11:37.462
and talk on their own, interesting[br]processes and productivities can emerge.

0:11:37.462,0:11:40.403
And now I think next time[br]you go to your front step

0:11:40.403,0:11:42.914
and pick up that box that[br]you just ordered online,

0:11:42.914,0:11:45.138
you break it open and[br]the goo is in there,

0:11:45.138,0:11:47.891
you'll have some wonderment[br]as to whether a robot

0:11:47.891,0:11:50.484
assisted in the picking[br]and packing of that order.

0:11:50.484,0:11:52.163
Thank you.

0:11:52.163,0:11:56.793
(Applause)