1
00:00:00,000 --> 00:00:03,561
I want to talk to you about,

2
00:00:03,561 --> 00:00:07,122
or share with you, a
breakthrough new approach

3
00:00:07,122 --> 00:00:10,464
for managing items of
inventory inside of a warehouse.

4
00:00:10,464 --> 00:00:13,022
We're talking about a pick,
pack and ship setting here.

5
00:00:13,022 --> 00:00:15,845
So as a hint,

6
00:00:15,845 --> 00:00:19,998
this solution involves
hundreds of mobile robots,

7
00:00:19,998 --> 00:00:22,422
sometimes thousands
of mobile robots,

8
00:00:22,422 --> 00:00:25,020
moving around a warehouse.
And I'll get to the solution.

9
00:00:25,020 --> 00:00:26,240
But for a moment, just think

10
00:00:26,240 --> 00:00:28,766
about the last time that
you ordered something online.

11
00:00:28,766 --> 00:00:30,830
You were sitting
on your couch

12
00:00:30,830 --> 00:00:34,890
and you decided that you
absolutely had to have this red t-shirt.

13
00:00:34,890 --> 00:00:37,186
So — click! — you put it
into your shopping cart.

14
00:00:37,186 --> 00:00:39,084
And then you decided
that green pair of pants

15
00:00:39,084 --> 00:00:40,982
looks pretty good too — click!

16
00:00:40,982 --> 00:00:42,882
And maybe a blue
pair of shoes — click!

17
00:00:42,882 --> 00:00:44,932
So at this point you've
assembled your order.

18
00:00:44,932 --> 00:00:46,700
You didn't stop to think 
for a moment that

19
00:00:46,700 --> 00:00:48,468
that might not be a great outfit.

20
00:00:48,468 --> 00:00:50,238
But you hit
"submit order."

21
00:00:50,238 --> 00:00:54,412
And two days later, this package
shows up on your doorstep.

22
00:00:54,412 --> 00:00:57,366
And you open the box and you're
like, wow, there's my goo.

23
00:00:57,366 --> 00:01:00,327
Did you ever stop to think about
how those items of inventory

24
00:01:00,327 --> 00:01:04,540
actually found their way inside
that box in the warehouse?

25
00:01:04,540 --> 00:01:08,358
So I'm here to tell you 
it's that guy right there.

26
00:01:08,358 --> 00:01:11,978
So deep in the
middle of that picture,

27
00:01:11,978 --> 00:01:14,493
you see a classic
pick-pack worker

28
00:01:14,493 --> 00:01:17,718
in a distribution or
order fulfillments setting.

29
00:01:17,718 --> 00:01:21,600
Classically these pick workers will 
spend 60 or 70 percent of their day

30
00:01:21,600 --> 00:01:23,512
wandering around
the warehouse.

31
00:01:23,512 --> 00:01:26,223
They'll often walk
as much as 5 or 10 miles

32
00:01:26,223 --> 00:01:28,884
in pursuit of
those items of inventory.

33
00:01:28,884 --> 00:01:33,305
Not only is this an
unproductive way to fill orders,

34
00:01:33,305 --> 00:01:37,362
it also turns out to be an
unfulfilling way to fill orders.

35
00:01:37,362 --> 00:01:40,993
So let me tell you where I
first bumped into this problem.

36
00:01:40,993 --> 00:01:45,047
I was out in the Bay area 
in '99, 2000, the dot com boom.

37
00:01:45,047 --> 00:01:49,109
I worked for a fabulously
spectacular flame-out called Webvan.

38
00:01:49,109 --> 00:01:50,930
(Laughter)

39
00:01:50,930 --> 00:01:53,660
This company raised hundreds of
millions of dollars with the notion that

40
00:01:53,660 --> 00:01:56,442
we will deliver
grocery orders online.

41
00:01:56,442 --> 00:02:00,615
And it really came down to the fact
that we couldn't do it cost effectively.

42
00:02:00,615 --> 00:02:04,577
Turns out e-commerce was something
that was very hard and very costly.

43
00:02:04,577 --> 00:02:08,828
In this particular instance we were trying
to assemble 30 items of inventory

44
00:02:08,828 --> 00:02:12,800
into a few totes, onto a van
to deliver to the home.

45
00:02:12,800 --> 00:02:16,600
And when you think about it,
it was costing us 30 dollars.

46
00:02:16,600 --> 00:02:19,686
Imagine, we had an
89¢ can of soup

47
00:02:19,686 --> 00:02:23,362
that was costing us one dollar to
pick and pack into that tote.

48
00:02:23,362 --> 00:02:26,570
And that's before we actually
tried to deliver it to the home.

49
00:02:26,570 --> 00:02:29,357
So long story short,
during my one year at Webvan,

50
00:02:29,357 --> 00:02:32,593
what I realized by talking to 
all the material-handling providers

51
00:02:32,593 --> 00:02:37,231
was that there was no solution designed
specifically to solve each base picking.

52
00:02:37,231 --> 00:02:41,254
Red item, green, blue, getting
those three things in a box.

53
00:02:41,254 --> 00:02:44,239
So we said, there's just
got to be a better way to do this.

54
00:02:44,239 --> 00:02:46,766
Existing material handling
was set up to pump

55
00:02:46,766 --> 00:02:50,533
pallets and cases of
goo to retail stores.

56
00:02:50,533 --> 00:02:54,330
Of course Webvan went out of business,
and about a year and a half later,

57
00:02:54,330 --> 00:02:57,910
I was still noodling on this problem.
It was still nagging at me.

58
00:02:57,910 --> 00:02:59,778
And I started
thinking about it again.

59
00:02:59,778 --> 00:03:04,905
And I said, let me just focus briefly
on what I wanted as a pick worker,

60
00:03:04,905 --> 00:03:07,236
or my vision for
how it should work.

61
00:03:07,236 --> 00:03:08,650
(Laughter)

62
00:03:08,650 --> 00:03:10,572
I said, let's focus
on the problem.

63
00:03:10,572 --> 00:03:14,427
I have an order here and what
I want to do is I want to put

64
00:03:14,427 --> 00:03:16,647
red, green and blue
in this box right here.

65
00:03:16,647 --> 00:03:19,499
What I need is a system where I put out
my hand and — poof! —

66
00:03:19,499 --> 00:03:22,022
the product shows up 
and I pack it into the order,

67
00:03:22,022 --> 00:03:23,715
and now we're thinking,

68
00:03:23,715 --> 00:03:27,729
this would be a very operator-centric
approach to solving the problem.

69
00:03:27,729 --> 00:03:31,956
This is what I need. What technology
is available to solve this problem?

70
00:03:31,956 --> 00:03:35,897
But as you can see, orders can come
and go, products can come and go.

71
00:03:35,897 --> 00:03:40,580
It allows us to focus on making the
pick worker the center of the problem,

72
00:03:40,580 --> 00:03:45,204
and providing them the tools to make
them as productive as possible.

73
00:03:45,204 --> 00:03:47,304
So how did I
arrive at this notion?

74
00:03:47,304 --> 00:03:50,837
Well, actually it came from
a brainstorming exercise,

75
00:03:50,837 --> 00:03:53,677
probably a technique
that many of you use,

76
00:03:53,677 --> 00:03:55,833
It's this notion of
testing your ideas.

77
00:03:55,833 --> 00:03:57,563
Take a blank sheet, of course,

78
00:03:57,563 --> 00:04:01,593
but then test your ideas
at the limits — infinity, zero.

79
00:04:01,593 --> 00:04:04,775
In this particular case, we
challenged ourselves with the idea:

80
00:04:04,775 --> 00:04:07,751
What if we had to build a
distribution center in China,

81
00:04:07,751 --> 00:04:10,365
where it's a very,
very low-cost market?

82
00:04:10,365 --> 00:04:13,578
And say, labor is cheap,
land is cheap.

83
00:04:13,578 --> 00:04:15,224
And we said specifically,

84
00:04:15,224 --> 00:04:18,230
"What if it was zero dollars
an hour for direct labor

85
00:04:18,230 --> 00:04:20,946
and we could build a million-
square-foot distribution center?"

86
00:04:20,946 --> 00:04:23,021
So naturally that
led to ideas that said,

87
00:04:23,021 --> 00:04:24,829
"Let's put lots of people
in the warehouse."

88
00:04:24,829 --> 00:04:27,374
And I said, "Hold on,
zero dollars per hour,

89
00:04:27,374 --> 00:04:30,231
what I would do is 'hire'

90
00:04:30,231 --> 00:04:34,358
10,000 workers to come to the
warehouse every morning at 8 a.m.,

91
00:04:34,358 --> 00:04:37,477
walk into the warehouse and
pick up one item of inventory

92
00:04:37,477 --> 00:04:39,210
and then just stand there.

93
00:04:39,210 --> 00:04:41,749
So you hold Captain Crunch,
you hold the Mountain Dew,

94
00:04:41,749 --> 00:04:43,172
you hold the Diet Coke.

95
00:04:43,172 --> 00:04:45,428
If I need it, I'll call you,
otherwise just stand there.

96
00:04:45,428 --> 00:04:48,815
But when I need Diet Coke and I call it,
you guys talk amongst yourselves.

97
00:04:48,815 --> 00:04:52,936
Diet Coke walks up to the front —
pick it, put it in the tote, away it goes."

98
00:04:52,936 --> 00:04:57,588
Wow, what if the products
could walk and talk on their own?

99
00:04:57,588 --> 00:04:59,926
That's a very interesting,
very powerful way

100
00:04:59,926 --> 00:05:02,520
that we could potentially
organize this warehouse.

101
00:05:02,520 --> 00:05:04,868
So of course,
labor isn't free,

102
00:05:04,868 --> 00:05:08,176
on that practical
versus awesome spectrum.

103
00:05:08,176 --> 00:05:09,866
(Laughter)

104
00:05:09,866 --> 00:05:12,983
So we said mobile shelving —
We'll put them on mobile shelving.

105
00:05:12,983 --> 00:05:17,776
We'll use mobile robots and
we'll move the inventory around.

106
00:05:17,776 --> 00:05:22,194
And so we got underway on that and
then I'm sitting on my couch in 2008.

107
00:05:22,194 --> 00:05:26,013
Did any of you see the Beijing
Olympics, the opening ceremonies?

108
00:05:26,013 --> 00:05:29,054
I about fell out of my
couch when I saw this.

109
00:05:29,054 --> 00:05:30,527
I'm like, that was the idea!

110
00:05:30,527 --> 00:05:35,026
(Laughter and Applause)

111
00:05:35,026 --> 00:05:38,917
We'll put thousands of people on
the warehouse floor, the stadium floor.

112
00:05:38,917 --> 00:05:42,952
But interestingly enough, this
actually relates to the idea

113
00:05:42,952 --> 00:05:48,364
in that these guys were creating some
incredibly powerful, impressive digital art,

114
00:05:48,364 --> 00:05:50,440
all without computers,
I'm told,

115
00:05:50,440 --> 00:05:52,886
it was all peer-to-peer
coordination and communication.

116
00:05:52,886 --> 00:05:54,462
You stand up,
I'll squat down.

117
00:05:54,462 --> 00:05:56,107
And they made
some fabulous art.

118
00:05:56,107 --> 00:05:58,795
It speaks to the
power of emergence

119
00:05:58,795 --> 00:06:02,763
in systems when you let things
start to talk with each other.

120
00:06:02,763 --> 00:06:06,541
So that was a little
bit of the journey.

121
00:06:06,541 --> 00:06:10,579
So of course, now what became
the practical reality of this idea?

122
00:06:10,579 --> 00:06:12,385
Here is a warehouse.

123
00:06:12,385 --> 00:06:16,161
It's a pick, pack and ship center
that has about 10,000 different SKUs.

124
00:06:16,161 --> 00:06:20,298
We'll call them red pens,
green pens, yellow Post-It Notes.

125
00:06:20,298 --> 00:06:23,677
We send the little orange robots
out to pick up the blue shelving pods.

126
00:06:23,677 --> 00:06:25,845
And we deliver them
to the side of the building.

127
00:06:25,845 --> 00:06:28,981
So all the pick workers now
get to stay on the perimeter.

128
00:06:28,981 --> 00:06:31,323
And the game here is
to pick up the shelves,

129
00:06:31,323 --> 00:06:34,647
take them down the highway and
deliver them straight to the pick worker.

130
00:06:34,647 --> 00:06:36,784
This pick worker's life
is completely different.

131
00:06:36,784 --> 00:06:40,201
Rather than wandering around
the warehouse, she gets to stay still

132
00:06:40,201 --> 00:06:41,789
in a pick station like this

133
00:06:41,789 --> 00:06:45,787
and every product in the
building can now come to her.

134
00:06:45,787 --> 00:06:49,065
So the process
is very productive.

135
00:06:49,065 --> 00:06:53,260
Reach in, pick an item,
scan the bar code, pack it out.

136
00:06:53,260 --> 00:06:54,909
By the time
you turn around,

137
00:06:54,909 --> 00:06:57,948
there's another product there
ready to be picked and packed.

138
00:06:57,948 --> 00:07:00,958
So what we've done is take
out all of the non-value added

139
00:07:00,958 --> 00:07:03,642
walking, searching,
wasting, waited time,

140
00:07:03,642 --> 00:07:07,920
and we've developed a very 
high-fidelity way to pick these orders,

141
00:07:07,920 --> 00:07:12,084
where you point at it with
a laser, scan the UPC barcode,

142
00:07:12,084 --> 00:07:15,278
and then indicate with a light
which box it needs to go into.

143
00:07:15,278 --> 00:07:18,764
So more productive, more
accurate and, it turns out,

144
00:07:18,764 --> 00:07:23,070
it's a more interesting office
environment for these pick workers.

145
00:07:23,070 --> 00:07:25,507
They actually complete
the whole order.

146
00:07:25,507 --> 00:07:28,034
So they do red, green and blue,
not just a part of the order.

147
00:07:28,034 --> 00:07:31,648
And they feel a little bit more
in control of their environment.

148
00:07:31,648 --> 00:07:34,846
So the side effects
of this approach

149
00:07:34,846 --> 00:07:36,414
are what really surprised us.

150
00:07:36,414 --> 00:07:37,992
We knew it was going 
to be more productive.

151
00:07:37,992 --> 00:07:42,062
But we didn't realize just how
pervasive this way of thinking

152
00:07:42,062 --> 00:07:47,292
extended to other
functions in the warehouse.

153
00:07:47,292 --> 00:07:52,124
But what effectively this approach
is doing inside of the DC

154
00:07:52,124 --> 00:07:56,588
is turning it into a massively
parallel processing engine.

155
00:07:56,588 --> 00:07:59,211
So this is again a
cross-fertilization of ideas.

156
00:07:59,211 --> 00:08:01,150
Here's a warehouse
and we're thinking about

157
00:08:01,150 --> 00:08:04,669
parallel processing
supercomputer architectures.

158
00:08:04,669 --> 00:08:07,066
The notion here is that you have

159
00:08:07,066 --> 00:08:09,863
10 workers on 
the right side of the screen

160
00:08:09,863 --> 00:08:14,111
that are now all independent
autonomous pick workers.

161
00:08:14,111 --> 00:08:18,015
If the worker in station three decides
to leave and go to the bathroom,

162
00:08:18,015 --> 00:08:21,499
it has no impact on the
productivity of the other nine workers.

163
00:08:21,499 --> 00:08:25,869
Contrast that, for a moment, with the
traditional method of using a conveyor.

164
00:08:25,869 --> 00:08:27,928
When one person
passes the order to you,

165
00:08:27,928 --> 00:08:30,427
you put something in
and pass it downstream.

166
00:08:30,427 --> 00:08:33,544
Everyone has to be in place
for that serial process to work.

167
00:08:33,544 --> 00:08:36,721
This becomes a more robust
way to think about the warehouse.

168
00:08:36,721 --> 00:08:41,265
And then underneath the hoods gets
interesting in that we're tracking

169
00:08:41,265 --> 00:08:43,003
the popularity
of the products.

170
00:08:43,003 --> 00:08:45,827
And we're using dynamic
and adaptive algorithms

171
00:08:45,827 --> 00:08:50,281
to tune the floor
of the warehouse.

172
00:08:50,281 --> 00:08:55,165
So what you see here potentially
the week leading up to Valentine's Day.

173
00:08:55,165 --> 00:08:59,008
All that pink chalky candy has
moved to the front of the building

174
00:08:59,008 --> 00:09:02,901
and is now being picked into a
lot of orders in those pick stations.

175
00:09:02,901 --> 00:09:07,064
Come in two days after Valentine's Day,
and that candy, the leftover candy,

176
00:09:07,064 --> 00:09:09,337
has all drifted to the
back of the warehouse

177
00:09:09,337 --> 00:09:13,630
and is occupying the cooler
zone on the thermal map there.

178
00:09:13,630 --> 00:09:17,113
One other side effect of this
approach using the parallel processing

179
00:09:17,113 --> 00:09:20,108
is these things can
scale to ginormous.

180
00:09:20,108 --> 00:09:21,635
(Laughter)

181
00:09:21,635 --> 00:09:24,369
So whether you're doing
two pick stations, 20 pick stations,

182
00:09:24,369 --> 00:09:27,642
or 200 pick stations, the
path planning algorithms

183
00:09:27,642 --> 00:09:30,185
and all of the inventory
algorithms just work.

184
00:09:30,185 --> 00:09:34,541
In this example you
see that the inventory

185
00:09:34,541 --> 00:09:36,867
has now occupied all the
perimeter of the building

186
00:09:36,867 --> 00:09:39,213
because that's where
the pick stations were.

187
00:09:39,213 --> 00:09:41,231
They sorted it
out for themselves.

188
00:09:41,231 --> 00:09:43,518
So I'll conclude with
just one final video

189
00:09:43,518 --> 00:09:46,502
that shows how
this comes to bear

190
00:09:46,502 --> 00:09:50,096
on the pick worker's actual
day in the life of.

191
00:09:50,096 --> 00:09:54,432
So as we mentioned, the process is
to move inventory along the highway

192
00:09:54,432 --> 00:09:57,084
and then find your way
into these pick stations.

193
00:09:57,084 --> 00:09:59,555
And our software in the background

194
00:09:59,555 --> 00:10:02,306
understands what's going on 
in each station,

195
00:10:02,306 --> 00:10:04,917
we direct the pods
across the highway

196
00:10:04,917 --> 00:10:07,631
and we're attempting to
get into a queuing system

197
00:10:07,631 --> 00:10:10,515
to present the work
to the pick worker.

198
00:10:10,515 --> 00:10:13,760
What's interesting is we can even
adapt the speed of the pick workers.

199
00:10:13,760 --> 00:10:17,686
The faster pickers get more pods
and the slower pickers get few.

200
00:10:17,686 --> 00:10:20,824
But this pick worker now is
literally having that experience

201
00:10:20,824 --> 00:10:22,677
that we described before.

202
00:10:22,677 --> 00:10:25,221
She puts out her hand.
The product jumps into it.

203
00:10:25,221 --> 00:10:27,426
Or she has to reach in and get it.

204
00:10:27,426 --> 00:10:29,871
She scans it and 
she puts it in the bucket.

205
00:10:29,871 --> 00:10:33,667
And all of the rest of the technology
is kind of behind the scenes.

206
00:10:33,667 --> 00:10:37,528
So she gets to now focus on the
picking and packing portion of her job.

207
00:10:37,528 --> 00:10:40,802
Never has any idle time,
never has to leave her mat.

208
00:10:40,802 --> 00:10:44,942
And actually we think
not only a more productive

209
00:10:44,942 --> 00:10:48,212
and more accurate
way to fill orders.

210
00:10:48,212 --> 00:10:51,622
We think it's a more
fulfilling way to fill orders.

211
00:10:51,622 --> 00:10:54,830
The reason we can say
that, though, is that workers

212
00:10:54,830 --> 00:10:56,972
in a lot of these
buildings now compete

213
00:10:56,972 --> 00:11:00,154
for the privilege of working
in the Kiva zone that day.

214
00:11:00,154 --> 00:11:02,817
And sometimes we'll catch
them on testimonial videos

215
00:11:02,817 --> 00:11:04,934
saying such things as,

216
00:11:04,934 --> 00:11:09,180
they have more energy after the
day to play with their grandchildren,

217
00:11:09,180 --> 00:11:13,776
or in one case a guy said, "the
Kiva zone is so stress-free

218
00:11:13,776 --> 00:11:16,891
that I've actually stopped taking
my blood pressure medication."

219
00:11:16,891 --> 00:11:18,725
(Laughter)

220
00:11:18,725 --> 00:11:22,724
That was at a pharmaceutical distributor,
so they told us not to use that video.

221
00:11:22,724 --> 00:11:26,292
(Laughter)

222
00:11:26,292 --> 00:11:29,316
So what I wanted to leave you
with today is the notion that

223
00:11:29,316 --> 00:11:32,171
when you let things start
to think and walk

224
00:11:32,171 --> 00:11:37,462
and talk on their own, interesting
processes and productivities can emerge.

225
00:11:37,462 --> 00:11:40,403
And now I think next time
you go to your front step

226
00:11:40,403 --> 00:11:42,914
and pick up that box that
you just ordered online,

227
00:11:42,914 --> 00:11:45,138
you break it open and
the goo is in there,

228
00:11:45,138 --> 00:11:47,891
you'll have some wonderment
as to whether a robot

229
00:11:47,891 --> 00:11:50,484
assisted in the picking
and packing of that order.

230
00:11:50,484 --> 00:11:52,163
Thank you.

231
00:11:52,163 --> 00:11:56,793
(Applause)