AWS re:Invent 2022 - Building the industrial kitchen of the future with Miso Robotics (ROB201)
AWS re:Invent 2022 - Building the industrial kitchen of the future with Miso Robotics (ROB201)
Miso Robotics is transforming the restaurant industry with robotics and intelligent automation with the mission to enhance customer satisfaction. In this session, learn how Miso Robotics and AWS collaborated on cloud-based testing with simulation, data warehousing and analytics, edge management, and more. Miso Robotics shares how they use AWS tools and technologies to understand, analyze, and improve the performance of Miso’s Flippy 2, Flippy Lite, and CookRight robots.
Learn more about AWS re:Invent at https://go.aws/3ikK4dD.
Subscribe:
More AWS videos http://bit.ly/2O3zS75
More AWS events videos http://bit.ly/316g9t4
ABOUT AWS
Amazon Web Services (AWS) hosts events, both online and in-person, bringing the cloud computing community together to connect, collaborate, and learn from AWS experts.
AWS is the world’s most comprehensive and broadly adopted cloud platform, offering over 200 fully featured services from data centers globally. Millions of customers—including the fastest-growing startups, largest enterprises, and leading government agencies—are using AWS to lower costs, become more agile, and innovate faster.
#reInvent2022 #AWSreInvent2022 #AWSEvents
Content
0 -> Hello and welcome.
2.751 -> So excited to have you all here.
4.71 -> Thank you for being here in advance
6.48 -> for your time.
7.71 -> I'm Deborah Matteliano.
8.94 -> I serve as AWS's global Head
11.28 -> for Restaurant Technology
13.44 -> and I'm passionate about the areas
15.63 -> where food and technology can collide.
18.15 -> And we're gonna be talking about
19.56 -> one of those areas today.
21.45 -> Chris Kruger from Miso Robotics
23.52 -> is going to introduce you into new ways
26.55 -> that his products are redefining
28.843 -> and pioneering the future of restaurants.
33.18 -> But before we talk about
34.98 -> the future of restaurants
37.701 -> let's take a quick look at the past.
39.69 -> So, raise your hand
41.67 -> if you've ever worked in a restaurant
43.95 -> or, if you know someone who has
46.65 -> or was impacted by the pandemic.
49.47 -> Yeah, I'm seeing a lot of hands.
51.51 -> So if, if you did know someone
53.76 -> or if you have worked in
the hospitality industry,
56.34 -> you're likely well aware
that the pandemic was
59.58 -> devastating for restaurants.
61.83 -> A hundred thousand
restaurants closed, in fact
64.59 -> and 95% of workers lost jobs or wages.
70.08 -> And then the challenges
haven't stopped there.
72.95 -> So when you think about
it, many of the customers
76.491 -> that Chris is going to
introduce you to today
78.99 -> weren't even sure whether
they'd be standing,
81.6 -> today, and even open and operating.
85.35 -> But, what has happened next
87.69 -> is that a whole new slew of challenges
90.12 -> have come for restaurants.
So, drive through
92.52 -> orders have spiked so much
94.77 -> and delivery orders have spiked so much.
97.95 -> And then you have that amidst inflation
101.16 -> rising food costs and a labor shortage.
104.22 -> In fact, there are 1.3
million job openings
108.42 -> within the hospitality industry right now.
111.24 -> So you might be asking
yourself at this moment,
113.04 -> well is that what this talk is about?
114.99 -> You know, are we going to fill those
116.55 -> job openings? Is this what robots
118.65 -> are going to help us do?
119.82 -> Are we gonna sort of
automate some of that work
122.7 -> or our robots here to make us all smarter?
125.7 -> And maybe some of that is true,
128.01 -> but really I wonder too are robots here
131.37 -> to help us be more creative?
134.19 -> So let's explore that possibility.
137.1 -> So the holy grail of
creativity is known as
140.22 -> the flow state. And this term was coined
144.614 -> by a Hungarian psychologist.
And he spent his
148.23 -> life studying creators
and artists to understand
152.16 -> what helped them create their best work.
156.28 -> And what was cool about his study is that
157.86 -> he found that there was
an optimal condition
160.47 -> for innovation. And that condition happens
164.01 -> when challenge, which
you see on the x-axis,
168.36 -> is just about equal to
the skill that you see on
171.365 -> the y-axis. So when you think about this
174.885 -> and you think about where restaurants were
178.05 -> in 2020 and even beyond,
they're all the way
181.47 -> to the right of that x-axis, right.
183.33 -> Like they're, they're totally immersed
185.46 -> in challenge and because there's maybe not
187.56 -> that level of skill and tools to handle
190.89 -> the big problems and
challenges they're facing
194.16 -> you end up in the anxiety state
195.87 -> at the lower quadrant of the graph.
198.725 -> And what Chris and his team have done
200.776 -> is automated and given
new skills and tools
202.98 -> to restaurants so they can get out of
204.96 -> the anxiety state and head right back into
207.69 -> the flow state. What
you're also gonna learn
210.085 -> and what's exciting about today, is that
212.656 -> the flow state is universal,
across restaurants,
216.99 -> operators, food makers who want to unleash
219.96 -> their creativity through
food into the world
223.08 -> and also amongst roboticist and engineers
226.83 -> who are also creating new products
229.26 -> in high stakes environment
that take great deals
232.47 -> of skill and are also
extremely challenging.
235.8 -> So I'm gonna hand it over to Chris, now.
237.81 -> He's gonna tell you how
his team of roboticists
240.99 -> has automated restaurants
to help them get back
243.84 -> into the flow state and
how it happened on AWS.
250.17 -> - Thanks Deborah, and
thanks for coming today.
252.45 -> It's great to have a live audience.
254.49 -> I'm Chris Kruger, I'm
the CTO at Misa Robotics.
257.94 -> We're a hundred person
startup, based in Pasadena
261.66 -> and our products range, kind of a gambit,
265.62 -> but really all focused on, restaurant tech
268.71 -> and especially kitchens, to automate tasks
272.01 -> but also to assist on other tasks.
274.89 -> And well, I've only been
at Miso for a short time
278.61 -> but I have a long history as a maker.
281.67 -> Prior to being at Miso, I was with
283.427 -> iRobot for several years, where I ran the
286.77 -> Roomba and Braava software teams
289.842 -> and, really got to know
some of the challenges
291.72 -> and robotics there and
prior to that many years
294.6 -> doing smartphones and connected devices.
298.44 -> And, really one of the things that,
301.14 -> I see is, that connects
us, is this flow state.
305.25 -> You know, we want to get to that place
307.35 -> where we can all do our best work
309.93 -> and, well, I'll talk
about that a little more,
313.89 -> but why don't we meet flippy.
316.5 -> And you can see a little
bit of the product
319.02 -> offerings that we have from Miso Robotics
326.127 -> (upbeat-inspiring music playing)
328.046 -> - Suite of Robotic Quick
Serve Restaurant Solutions.
330.45 -> We design them to automate manually
332.7 -> repetitive back of house tasks.
335.88 -> For example, our partners
are currently piloting
338.58 -> flippy 2 for tasks like perfectly cooking
341.28 -> crispy fries, wings, mini tacos and more.
345.63 -> And they can use the
entire suite in the future.
348.36 -> That means frying up fresh tortilla chips,
351.36 -> cooking and saucing wings,
precisely monitoring
354.57 -> food temperatures, starting
with CookRight coffee
357.63 -> and filling drinks, all working together
360.15 -> in an automated market.
That's opportunity!
363.99 -> And that's where you come in. Miso is
366.42 -> looking for investors. When you invest,
368.88 -> you can be a part of our journey to
370.62 -> create a safer, easier, and well,
372.958 -> friendlier world of
quick serve restaurants.
376.152 -> (upbeat-inspiring music stops)
378.33 -> - Okay, well we are a crowdfunded company,
381.03 -> but our round is closed right now.
382.38 -> So this is not an
investment talk, (chuckling)
385.74 -> but you got a chance to see Flippy
388.05 -> and we'll show a lot more about him,
389.97 -> but also Chippy, we'll talk a little bit
392.79 -> about our CookRight product line,
395.19 -> which is computer vision and IoT sensors
398.49 -> for monitoring things like the performance
401.25 -> on a coffee station or at a grill.
404.73 -> And finally Sippy, our
automated drink dispenser
407.7 -> which should launch next year.
410.28 -> So, these are some of our products
412.98 -> that we're developing
to help kitchens really
417.78 -> find their flow state. Let's see how
419.94 -> that really works with flippy.
424.922 -> \ So this video here was a Time series
430.658 -> \ that we did
431.866 -> \ with an early prototype of the product.
433.784 -> \ And what it's meant
to show is on one side
435.507 -> \ you can see what our
associate Jake would look like
440.391 -> \ if he were a fry cook
with and without flippy.
445.302 -> \ So with flippy, what you can see is that
448.518 -> \ most of the items come from an automated
451.748 -> \ refrigerator that's
there, I believe on well,
452.581 -> \ it's off to the right on my side
456.957 -> \ and it's your high
volume items are gonna be
461.054 -> \ in there and can
automatically be dispensed
463.887 -> \ however you'd like.
And then the lower volume
467.884 -> \ items come from the
white bins in the front.
470.91 -> \ And basically you can take food
472.63 -> \ and place it in there
with computer vision.
475.122 -> \ It will recognize the food
476.942 -> \ and look up the recipe for it.
479.248 -> \ And what this does is it allows more
481.706 -> \ food to get cooked but most of the,
484.434 -> \ the labor in the fry station is removed
487.542 -> \ and this frees up the associate to
490.475 -> \ also help process
orders and help some of
493.428 -> \ that flow in the back of the kitchen.
501.78 -> \ We'll let him finish here.
511.82 -> \ And I think the main
thing that I take away
514.583 -> \ from any time I spend
in the kitchen is that
517.32 -> the kitchen environments
are very challenging.
519.21 -> So anything you can do
to help free up labor
522.223 -> and increased flow is
really key to restaurants.
526.89 -> And you know, while it's challenging
528.96 -> in the back of the kitchen,
it's also challenging
531.33 -> making a product like Flippy.
534.87 -> So we look at some of the challenges
536.37 -> that we face.
537.93 -> Really to make a product like Flippy,
541.11 -> takes dozens of technologies
to be integrated together
544.8 -> and to work in real time.
546.39 -> Many open source libraries, as well as,
548.64 -> Bespoke software, is brought together
551.4 -> and has to work in a real time environment
554.79 -> and with high reliability.
556.89 -> Many of the restaurants
we go in are 24 hour
560.01 -> restaurants seven days a week.
562.02 -> So they don't like downtime.
564.14 -> And we need, we have a lot of, kind of,
567.6 -> high level needs. So
we have over a gigabyte
570.21 -> of data that streams out of the devices
572.37 -> each hour. So as we bring those together
576.39 -> in the cloud that can be a challenge.
578.58 -> And a lot of the data is
time sensitive, as well.
582.66 -> And we need to analyze
this data, organize it,
585.54 -> store it, cause a lot of our big brands
588.15 -> also want us to know, wanna know
590.91 -> how well the the product's performing
593.7 -> but also how well their
restaurants are performing
597.06 -> And we build a lot of
this with scale in mind
601.17 -> because we're working with big brands
603.09 -> and they've got lots of
locations they'd like
605.58 -> us to scale to, as well
as, different regions.
609.75 -> We've got markets in Europe and Asia
614.04 -> opening up, as well.
615.96 -> And finally, all of
this has to be reusable
619.44 -> because things that we
build on one product
621.93 -> we wanna make sure that
we can use someplace else.
624.27 -> Cause as a small business,
we wanna make sure
626.4 -> that we're maximizing
our development dollars.
628.95 -> So the data transports
that we're using on Flippy
631.98 -> we also use for CookRight Coffee
634.41 -> the software update
mechanisms that we're using
638.25 -> on CookRight Coffee we
also want to use on Sippy.
642.54 -> So we wanna make sure
that those components,
644.73 -> you know, once we've got them hardened
646.59 -> are also available to
our other product lines.
649.8 -> And finally, simulation.
651.69 -> Simulation is really a key part
653.58 -> of modern robotics because
testing robot software
658.17 -> on hardware can be very expensive.
661.08 -> And not to mention many
of our installations
663.96 -> are custom. So while we have a nice big
666.87 -> commercial kitchen in Pasadena
668.49 -> we don't have a version of all
670.32 -> of our products there
all the time to test on.
672.99 -> So we've built sim worlds
and test them there.
677.64 -> And this also provides faster turnaround
682.02 -> on changes because with sim world you can
685.98 -> well, let me just talk
about that more directly.
691.86 -> Yeah, we create a sim world
694.05 -> for every one of our Flippies
695.94 -> and then we're able to
test software changes
700.35 -> on all of those immediately.
703.14 -> So using AWS robomaker
705.78 -> we put that simulation up into the cloud
708.51 -> and we can kick off all those simulations
710.31 -> in parallel. That way
we can check one change
713.55 -> across a whole fleet within a few minutes.
716.58 -> This is something that we did at iRobot
719.31 -> and it dramatically improved the velocity
722.82 -> of the developers. At
iRobot, we were basically
725.94 -> able to test one software change
727.89 -> across the whole fleet of robots.
729.72 -> We had a single software
baseline that went
731.58 -> across many different products.
733.56 -> So not every developer
could have every product
736.17 -> to test on, but once you threw it up
738.3 -> into the cloud, they were able
739.74 -> to get that feedback.
741.42 -> At iRobot, we were able
to do about 40 hours
743.73 -> of testing in 30 minutes, right.
746.01 -> By kicking off multiple sims at one time.
749.52 -> That kind of feedback
helps developers really
753.03 -> have confidence to move faster.
755.58 -> And at Miso what we're
doing is making sure
759.42 -> that each of those locations stays working
761.97 -> with each of the releases we do.
764.16 -> Now, once you have a performance
simulation environment
767.79 -> like we've developed for validation
770.7 -> we can then start to use that environment
772.68 -> for other things. We get
confidence in that software
776.19 -> and we can then try out
performance improvements on it.
779.76 -> Or in the case of Chipotle,
when they came to us
783.57 -> and asked us if we could customize flippy
785.58 -> for them to cook their chips, we looked at
788.37 -> several different variations and decided
790.92 -> from sim results that we needed to build
793.92 -> kind of a new robot for them.
795.87 -> So we created Chippy, a single use
798.6 -> or a single product, chip cooking robot
803.97 -> that also limes and salts in.
808.718 -> And what was quite
amazing is when they came
811.68 -> to our lab and saw the units they looked
814.02 -> very much like what we proposed to them
815.82 -> because honestly we had built it out
817.89 -> before we showed them
but only in simulation
821.22 -> and they were really able to fine tune
823.2 -> the experience that came out of it.
825.09 -> So Chef Neville came to our office
827.16 -> and tested chips for the day and he,
830.1 -> came back and he said, they're great
832.41 -> but they're all too uniform. And that's
835.05 -> not really how we do it at Chipotle.
837.3 -> At Chipotle, you know,
the chips have kind of
839.67 -> a natural variation.
Some have too much lime,
841.95 -> some a little too much
salt and we'd like it
844.26 -> to be more like the signature experience.
847.31 -> So we actually added that variability
849.3 -> into the robot until we got to a point
851.31 -> where it really matched what they wanted
854.85 -> basically put more of the human
856.95 -> back into the robot.
859.08 -> And a lot of this is
enabled by our ability
861.45 -> to sim that out first before we build it.
865.89 -> And simulation was really one of the first
868.77 -> places that I got to know AWS services
873.45 -> and kind of found out
like what a great partner
877.02 -> they can be.
880.32 -> And so when we needed cloud resources
883.56 -> and cloud solutions, we came to AWS
888.33 -> and they combined us, or they introduced
890.76 -> us to Eplexity one of their partners.
893.55 -> And Eplexity was able to provide us with
896.97 -> a proposal to move all of our data
900.45 -> to a single architecture and to build
904.71 -> out what ended up being a data lake
906.99 -> for us with some other
enhanced data capabilities
911.28 -> in the fraction of the time that we
913.89 -> would've taken to build
it. But more importantly,
917.22 -> they were able to utilize
some of the services
920.207 -> AWS that they were
familiar with like the AWS
924.93 -> Lake Foundation. And
that brought in elements
929.1 -> that we were a bit
behind on, like security
932.4 -> access control, and
separating out the data
936.12 -> between some of the big
brands that we service
940.08 -> and with these capabilities and another
943.17 -> one that was really
great, was the AWS glue.
946.77 -> They were able to give
us a solution that kept
949.59 -> up with our changing schema.
951.45 -> We're developing a lot of products
953.28 -> and there was a lot of change and we
955.08 -> were spending a lot of
time keeping those in sync.
958.383 -> AWS Glue was able to do
that almost automatically.
962.1 -> And what it ended up
doing was providing us
964.53 -> really, the the big data
platform that we needed
967.5 -> so that we could use Athena and really get
970.62 -> to the data in ways
that we were struggling
972.38 -> to do before. So a lot of our developers
975.93 -> now have a rich playground
to go investigate
978.713 -> different aspects of the data that weren't
980.811 -> as easily accessible,
but we're also able to
983.76 -> then put together
production reports for our
986.1 -> customers and let them
know how their restaurants
988.98 -> and our products are doing
kind of in real time.
996.33 -> We also layered on top
some real time analytics
1000.5 -> because as I mentioned,
as our restaurants are
1003.679 -> many of them are 24-7.
They also, in peak periods
1007.79 -> they don't want a piece of equipment
1009.971 -> to ever go down. And if it does,
1011.09 -> they want it to recover very quickly.
1014.24 -> So by utilizing the AWS greengrass client,
1018.86 -> as well as IoT core, we
were able to stream data
1022.49 -> and get some live data
streams on the pieces
1024.62 -> of data that are important to
1026.63 -> really our support agents so that they
1028.85 -> could find and fix problems
almost immediately.
1033.62 -> It really was a game changer for us
1036.74 -> in shortening that response
time for the support agents
1040.34 -> and really helped focus them
1041.81 -> right where they needed to be.
1043.85 -> And all of these components were easier
1046.79 -> for us to build in
because AWS has a number
1049.54 -> of great examples, as well as, some native
1052.49 -> integration with Ross the
robot operating system
1056.06 -> which many modern robotics applications
1059.27 -> including Flippy and
our CookRight products
1062.57 -> are based on. And this
is really enabling a lot
1066.29 -> of what's out there and
what we wanna do do is
1069.14 -> we wanna stay focused and focus our really
1072.211 -> limited set of resources of developers
1075.92 -> and roboticists on that part of
1078.541 -> the development process. We don't really,
1081.47 -> you know, data lakes and
data models in general
1085.58 -> are not our core business.
They're an enabler
1087.98 -> to our core business.
And so, these components
1091.13 -> really helped us get there much quicker
1093.41 -> than we would've done on our own.
1097.79 -> And as we know, AWS
has new services coming
1100.22 -> out all the time and they
already have a number
1102.56 -> of services that we'd like to grow
1104 -> into and we've roadmapped,
places where like
1106.368 -> taking our existing
machine learning models
1109.85 -> and moving those to the cloud
1111.68 -> and federating that
data, kind of consumption
1114.83 -> better, are all things
that we'd like to do
1117.43 -> as we move forward. And
it's one of the things
1120.44 -> we actually did at iRobot,
where we put the user
1123.32 -> in the loop and whenever we would come
1126.14 -> across an item with like the j7 product
1129.68 -> it would ask the user if
they should clean that
1132.02 -> or avoid it to know whether
they should get stuck
1135.11 -> on something or whether they should really
1137.42 -> dig in and clean. And that those learnings
1140.66 -> are then federated back up to the cloud
1142.79 -> to make the data models or the data sets
1146 -> more rich and then all the products
1148.7 -> learn from that.
1150.17 -> Now that's kind of, you
know, on our roadmap in
1152.877 -> in the future and we know
that AWS has services
1155.12 -> that will handle that for us.
1161.36 -> And this is really why we do it.
1163.79 -> Building that loyalty with the customers.
1167.87 -> Diana from the White
Castle in Lokenia Illinois
1173.48 -> said, please don't take my Flippy away.
1175.64 -> I mean that's the kind
of satisfaction that
1178.49 -> really drives all of us to make
1180.02 -> the things we're doing. This product is
1182.6 -> helping them. It's alleviating pain points
1186.62 -> in the back of the the
restaurant for them.
1189.14 -> It shows up every day, it cooks the food
1191.78 -> perfectly every time and the team members
1195.32 -> really get to, rely on it and count on it.
1198.56 -> And it's not just White
Castle, it's a number of
1201.98 -> big brands have kind of
discovered the products
1204.86 -> that we're working on and
they're in early stages
1208.91 -> of trying those out.
And we're very excited
1212.09 -> about the growth future for Flippy
1215.9 -> and all of the Miso
products. And we're actually,
1220.34 -> very excited to also go global.
1232.82 -> Okay, so I guess this
one won't play the audio,
1236.81 -> but, we have.
1238.338 -> (technical difficulty downing out speaker)
1240.71 -> Or maybe it will
1242.125 -> - Miso Suites, Back of house solutions.
1242.958 -> Restaurants can easily
create more consistency
1245.78 -> with a faster speed of service
1247.4 -> all with the same great taste.
1249.5 -> It's the kitchen of the
future and it's here now.
1255.8 -> - So we have recently gone live in the UK
1260.797 -> and in a couple of weeks we will
1263.15 -> go live in Dubai. So
we, really are kind of
1266.99 -> expanding our footprint and we expect that
1270.515 -> in the future, it won't be just flippy,
1273.29 -> but there'll be a number of products
1274.123 -> that'll be connected in
the back of the kitchen
1277.13 -> that'll be able to serve a number
1279.83 -> of different functions
and talk to each other
1284.21 -> and figure out how to best
serve the restaurants.
1287.69 -> And that's our architecture
for the kitchen of the future.
1293.27 -> Thank you.
1295.407 -> (Audience clapping)
1300.29 -> - Thank you Chris.
1301.28 -> Thank you so much for being here,
1302.87 -> for sharing your story,
how you're innovating
1305.03 -> on behalf of restaurant customers.
1307.52 -> And Chris shared a whole lot about
1309.83 -> what AWS can do for roboticists
1312.155 -> but we don't want to limit you to
1314.36 -> what Chris shared. We wanna give you
1315.92 -> a picture of the full landscape and the
1318.137 -> the playground that's ripe for innovation
1321.26 -> and simulation. So some
of the other capabilities
1324.65 -> that we have are storage
database analytics
1329.09 -> as well as, like Chris said,
the ability to simulate.
1333.38 -> And at AWS we are always reinventing
1337.01 -> on behalf of our customers.
1339.05 -> One of our RoboMaker customers is actually
1342.2 -> amazon.com. So Amazon uses RoboMaker to
1345.861 -> simulate different scenarios and predict
1349.34 -> demand so that we can consistently meet
1352.43 -> the promises for delivery that we make
1354.71 -> to our customers every day.
1357.68 -> So now that you've
learned a little bit more
1359.84 -> about our robotics landscape
and how you can play
1362.87 -> and get into flow with
what we have to offer
1366.02 -> we'd also like to make a couple
1367.7 -> of recommendations for
how you can spend some
1369.957 -> of your additional time at re:Invent
1372.89 -> and make the most of it. One session
1375.47 -> that I'd like to call out would be
1377.66 -> Taco Bell session. This one's gonna be
1379.97 -> fascinating because
Taco Bell is pioneering
1383.27 -> new ways to use ML and
forecasting to optimize
1387.35 -> the third party delivery
ordering experience
1390.83 -> and their understanding
of what's happening
1393.44 -> across that chain. So
definitely check that out.
1396.23 -> Don't miss Taco Bell!
1398.93 -> And to wrap up,
1400.46 -> we wanna say thank you
again. We're so grateful
1402.98 -> for your time.
Source: https://www.youtube.com/watch?v=a2UaDLjZs_I