Episode Transcript
Transcripts are displayed as originally observed. Some content, including advertisements may have changed.
Use Ctrl + F to search
0:00
Hi.
0:01
This is Graham Class, producer of the
0:03
Daily Dad Jokes podcast. You may
0:05
remember me from such episodes as Daily
0:08
Dad Jokes Explained Anyway,
0:11
coming up as an episode from my other
0:13
podcast, Technically Speaking and
0:15
Intel Podcast. I hope you enjoy
0:18
it. But before I go, here's a
0:20
dad joke I heard at the
0:22
local factory. The management is replacing
0:24
all their janitors with robots.
0:28
Talk about making sweeping changes.
0:35
Take a second to think about every single
0:38
item in your home, your television,
0:40
your refrigerator, your death lamp,
0:43
your laptop, even the smartphone
0:45
you might be using to hear my voice right
0:47
now. All of these things and
0:50
so many more items in our lives began
0:53
in a factory. There
0:55
are more than six hundred and twenty thousand manufacturing
0:58
businesses in the United States right
1:00
now, responsible for nearly twelve
1:02
percent of the total US economic
1:04
output. The numbers are even more
1:06
staggering in China, which makes up
1:09
nearly twenty nine percent of the total
1:11
global output. For manufacturing. Factories
1:14
have been around since the late eighteenth century,
1:17
and today they're used everywhere from
1:19
South Korea to southern California
1:21
to make cars, airplanes,
1:23
textiles, and even space vehicles
1:26
and each one depends on a carefully
1:29
choreographed system of steps, each
1:32
one as essential as the next before
1:34
the final product rolls off the production
1:37
line. Mistakes, however,
1:40
are also an unavoidable part of this process.
1:43
Manufacturers simply can't check every
1:45
piece of every product, and
1:48
it's nearly impossible to achieve perfection when
1:50
some manufacturing plants produce thousands
1:53
of items a day. So
1:55
how can technology help an industry so
1:57
crucial to our daily lives? How
1:59
canactories use AI to reduce
2:02
and even prevent defective products?
2:07
Welcome to Technically Speaking, an Intel
2:09
podcast produced by iHeartMedia's
2:12
Ruby Studio in partnership with Intel.
2:15
In every episode, we explore how
2:17
AI innovations are changing the world
2:19
and revolutionizing the way we live. Hey
2:23
there, I'm gram class, and
2:25
today we're headed into the world of manufacturing,
2:28
an expansive and essential industry that
2:30
drives the global economy and boast
2:32
the history dating back nearly two hundred
2:34
and fifty years, we've seen manufacturing
2:37
create a revolution, resurrect nation's
2:39
economies, connect people around the
2:41
globe, and even send mankind
2:44
into space. But what's
2:46
next at the intersection of manufacturing
2:48
and technology. In this episode,
2:51
we'll be focusing on how AI technology
2:53
can help optimize manufacturing and
2:55
improve quality thanks to no small
2:58
part to the minds at Intel and at our Innovations,
3:01
a company committed to helping organizations
3:03
unlock the power of machine vision to
3:05
automate quality inspections. Before
3:08
we go any further, let's welcome our guest
3:12
joining us today is John Weiss, the chief
3:14
revenue Officer at Eigen Innovations.
3:17
John oversees all revenue generation activities
3:19
at Eigen, including driving sales
3:21
in Eigen's machine vision software and engineering
3:24
services.
3:25
Welcome to the show, John, Thanks for having me, Graham,
3:27
it's great to be here.
3:32
Let's start with a bit of background on manufacturing
3:34
and the role it plays in our society.
3:36
I mean, it's fair to say that I phone, our car,
3:38
laptop, even the food we eat involves
3:41
some sort of manufacturing process. I'd
3:44
like to get your thoughts on just the importance
3:46
and scale of manufacturing plants around the world.
3:49
Yeah. Sure, Well, like you said, just
3:51
about everything in our daily lives comes
3:53
from factories or plants. But
3:56
sure, depending on if you commute
3:59
on a train or in a car,
4:01
lots of those components are coming from factories.
4:03
Very little these days are really kind
4:05
of handcrafted and handmade and small
4:08
batch, especially large scale
4:10
consumer items. And there's many
4:12
different types of processes and many
4:15
different types of ways things are made.
4:18
And look, I know there's a
4:20
multitude of ways and types
4:22
of manufacturing processes.
4:24
Like a volkswag And built in Germany is going
4:26
to be very different from an iPhone built in China.
4:30
Do you find common threads or
4:32
similarities across manufacturing
4:35
industries.
4:36
Yeah, definitely, so, I guess maybe
4:38
as a kind of a starting point, it's important to understand
4:41
there are two types of manufacturing processes
4:43
or approaches. One is called process manufacturing.
4:46
This is things like chemicals, plastics,
4:49
things that can't really be broken down
4:52
or deconstructed easily. Or
4:54
you have discrete manufacturing, which
4:56
is much more of the process of putting stuff together.
4:59
Think about a watch or a car. Now,
5:01
both of those processes, discrete and process
5:04
manufacturing, they're quite different,
5:06
but there are certainly similarities.
5:08
And between the two methods. You basically
5:10
have all of the things that we use every day,
5:13
right then oftentimes actually they kind of
5:15
bleed into one another. Most things have a little bit
5:17
of process manufacturing involved, and then a little bit
5:19
of discrete manufacturing as well. However,
5:22
I would say the commonalities across
5:24
both are really heavily reliant on technology.
5:27
We see a very large push for data
5:29
driven decision making. We
5:32
see large patterns or trends
5:34
in both realms of manufacturing around
5:36
empowering the workforce, trying
5:39
to upscale workers via technology
5:41
to get them to be focused
5:43
on more mission critical tasks or
5:46
higher value activities while
5:48
letting some of the technology do more of the mundane
5:51
tasks.
5:52
Yeah. In a previous job that I had, we
5:54
were doing consumer electronics, and
5:57
we struggle quite a bit with
5:59
the quality side of things, sure, and being
6:01
able to ensure a good product
6:03
can you manufactured in our China plant?
6:06
And one thing that struck me was
6:09
that there was a very manual process
6:11
in terms of the quality inspection, and
6:13
it'll take samples in a one out of every
6:16
ten and they'll test that and then if that worked
6:18
and okay, then we assume the rest kind
6:20
of work. Right. I'm wondering if you
6:22
could share any stories or examples
6:24
of I guess problems with quality
6:27
or defective products that stick in
6:29
your mind. Yeah.
6:31
Absolutely, I mean that's all we do at AIGEN, right,
6:33
all we do is industrial machine vision
6:35
for inline quality inspection. So
6:37
a couple that stick into my mind actually
6:39
very relevant to what you said, sample testing.
6:42
Lots of manufacturers do this right if there are a
6:44
high volume shop or a high volume
6:46
process. For example, we have some
6:48
customers that use our technology to inspect
6:50
upwards of forty thousand units a week per
6:53
facility. The challenge is if you do find
6:55
a problem, now you're kind of scratching your head
6:57
wondering how many in between the last
6:59
one hundred or last fifty also
7:01
had a problem right, And unfortunately
7:04
you tend to find out the hard way when
7:06
you get returns or warranty claims
7:08
that maybe something wasn't right in that process.
7:11
And so technology is a great
7:13
way, especially the arena that we operate in computer
7:15
vision, we're helping customers
7:17
actually get away from that. A great example
7:20
is one of our manufacturing customers
7:22
who makes fuel tanks for a variety
7:24
of different vehicles, and
7:27
they do what's called destructive testing. They
7:29
don't just test, they actually break the fuel
7:31
tank, that cut it up, and they look
7:33
at all of the plastic components inside
7:35
and they see was it molded correctly, was
7:37
it welded correctly? And if
7:39
they have a problem, well, now they have to reverse
7:41
engineer a whole bunch of stuff and try to figure out,
7:44
holy cow, what went wrong? Right? And how do
7:46
we ensure that no bad fuel tank gets
7:48
on a truck. They started the journey with
7:50
us about three years ago, and fast forward today,
7:52
we're builts back on every new machine
7:54
that gets put into those plants for fuel
7:56
tank inspection. So they know unequivocally
7:59
every single product that they ship
8:01
out the door is of the highest quality standard
8:04
and if it's not, if something happens, now
8:06
they have complete traceability on everything
8:08
they've made, so they can figure out exactly
8:11
what went wrong in the process.
8:14
What John is talking about here is the output
8:16
of the manufacturing process. How can
8:18
we ensure every fuel tank that leaves
8:21
the plant will work as designed? Just
8:23
as importantly, we need to consider the quality
8:26
of the input components. Everything from
8:28
the greater steel to the precision
8:30
of the fuel gauges. These need
8:32
to be inspected to ensure that these are
8:34
up to the manufacturers standard. I
8:37
asked John for his thoughts about this.
8:43
We don't often look at raw material,
8:45
although it's possible in some cases, but
8:48
more often than not, are inputs that
8:50
we're looking at. It's actually process inputs
8:52
or parameters. So we're looking at feed
8:55
rates of raw materials, temperatures
8:57
of raw materials, things like this
8:59
and thes that become more of a scientific
9:02
look of what's happening on the assembly line
9:05
and ensuring that everything is inspect
9:07
We don't just look at the output of
9:10
you know, did you make a good or bad product, but we'll
9:12
actually show you all of the process data that
9:14
went into making that product. The
9:16
other side is on the discrete world,
9:18
where you're actually assembling things. In
9:21
this instance, what we do is we'll actually
9:23
monitor the assembly. So we'll look
9:26
at how people are placing door
9:28
panels into a doorframe,
9:30
for example on an automotive asset,
9:32
or look at tail lamps for
9:35
lighting purposes right the way that they're assembled
9:37
and put together. And what we can
9:39
do in real time is tell folks, hey,
9:41
what you're putting together is misconfigured, or
9:44
it's missing components, or it has too
9:46
many components. Those are defect
9:48
types that are pretty common in the assembly world.
9:51
And what are some of the technology that
9:54
is used for that? Is it vision? Is it sensors,
9:56
is it combination?
9:58
Everything we do is vision based so
10:00
we don't make cameras. By the way, we are a software
10:03
provider, we also act as a system integrator,
10:05
so a large part of our business is actually
10:07
delivering turnkey solutions,
10:09
not just the software. But we don't
10:11
make hardware, which is actually really
10:13
cool for us because that means we get to use
10:16
tons of different types of options that
10:18
are available for our customers and it
10:20
helps us really find the perfect
10:23
design and configuration that is
10:25
definitely going to solve problems. And
10:27
so having the flexibility is really nice, and of
10:29
course that's a large reason why we
10:31
partner with Intel. We're
10:33
built on the open Vino tech stack, and that
10:36
means we can run our software
10:38
really on any device that leverages
10:40
an Intel chip, which gives us tons
10:43
of options for deployments. What's
10:45
really cool about this though, from a quality perspective,
10:48
is that it means you now have one vision system
10:50
that can integrate with different types of
10:52
sensors. So if you want to do
10:55
say an optical inspection
10:57
for surface defects like scratches and dents,
11:00
but you also want to look at perhaps
11:02
inside that product in a thermal application,
11:05
if it's a molded part or something like that,
11:07
well you can look at all those different types
11:09
of sensor in one easier to use
11:11
screen right, So it removes
11:13
the headache of having to have five six different
11:16
vision systems to do a variety of inspections.
11:18
And I'm also interested in
11:20
the deployment of these sorts of new
11:23
technologies. I'd like to get your
11:25
thoughts and experiences around what's some of
11:27
the tips and tricks for people out there
11:29
trying to deploy not just for manufacturing
11:32
quality, but technology and AI in
11:34
general into a workforce that
11:36
maybe is a little bit hesitant.
11:38
Yeah, humans don't like change, that's for sure.
11:41
I know I don't. I'm guilty of that. And it's
11:44
certainly like that when you go into a
11:46
factory and you've got folks that have been
11:48
on the same line or in the
11:50
same steel plant for
11:53
twenty five thirty years, and
11:55
you show up and you've got this bright, new shiny
11:58
software and you say, hey, don't worry, data
12:00
is going to solve everything. Naturally, people
12:02
can be quite apprehensive. We don't
12:05
often run into technology challenges
12:07
anymore.
12:07
Now.
12:08
It's really we run into people challenges
12:10
and organizational challenges. So
12:12
first and foremost, I'll give the advice that I give
12:14
on most of the times I'm asked this question, but it's
12:17
so true. Is you don't ever
12:19
start adopting technology just
12:21
for the sake of adopting it, just because
12:23
competitors are using something, or just because
12:25
somebody wigh up the chain says, hey, we need an AI strategy,
12:28
Go invest in AI boom. Spend
12:30
some time and really think about the problems
12:33
that you're trying to tackle. In my world,
12:35
in the quality world in manufacturing, it's
12:37
looking at things you can do to increase yields,
12:40
increase your throughput, reduce your waste,
12:43
reduce your rework, and ultimately
12:45
lower what's called the cost of quality.
12:48
Start with that, find a way that you
12:50
can or process that you can optimize
12:53
by using some of this newer technology, and then
12:55
of course do a cost assessment or
12:58
a return on your investment analysis, and
13:00
ensure that the business justification is there.
13:03
My experience, that's where a lot of these projects
13:05
fall short, and where folks get stuck in
13:07
these pilots and pocs is because
13:09
they get really excited to try something,
13:12
but there is no proven business
13:14
value or business justification behind
13:16
it, and naturally then you don't get
13:19
the executive sponsorship you need. Your budget
13:21
falls through and the project goes nowhere.
13:23
And in your experience, what industries do
13:25
you find actually a
13:28
little bit more advanced in terms of adopting
13:30
these new technologies both on a
13:32
technical level but also at an organizational
13:34
level that it seems like the teams
13:36
are actually involved and successfully
13:39
deploying these sorts of techniques.
13:41
Yeah, that's a great question. We see pretty
13:44
advanced deployments in the automotive world
13:46
as far as discrete manufacturing goes. They
13:48
tend to be far ahead of the curve compared
13:51
to say steel manufacturers
13:53
or something like that, or concrete manufacturers.
13:56
There's a lot of very advanced technology and
13:58
those automotive facilities that make sure what you
14:00
buy is actually perfect. Similarly,
14:03
in the process world, pharmaceuticals
14:05
tends to be on the continuous process side
14:07
that tends to be pretty advanced. They have
14:09
a lot of vision systems in place looking at
14:12
the vaccine vials to ensure
14:14
the integrity of vile caps and seals
14:17
and things like that. Some of the laggers
14:19
would be metals, some of the plastics
14:22
organizations. But there's also
14:24
a kind of a bigger dynamic in manufacturing that
14:26
I think folks don't really understand
14:28
that also contributes to who's advanced
14:31
and who's not, which is the
14:33
sheer size of these organizations,
14:35
right, Manufacturers are not all large. Folks
14:38
tend to think about John Deere and three M
14:40
and you know, the largest players
14:43
in the world, and the reality is that makes
14:45
up such a small fraction
14:48
of the manufacturing pool. Especially
14:50
in America, most manufacturing facilities
14:53
have you know, twenty people or less, small
14:56
to medium manufacturers anywhere
14:58
from say like the twenty to two hundred range employees.
15:01
That's who makes up the vast majority
15:03
of our products. Even when you buy something
15:05
really big, you know, whether it's a whirlpool dishwasher,
15:08
or a hot tub or whatever it might be, all
15:10
those little components that make up
15:13
that consumer good, Well, it came from
15:15
probably many different suppliers, and most
15:17
of those are small.
15:18
It's nice that you mentioned that because my father
15:20
has a small manufacturing facility
15:23
here. And just to talk a little
15:25
bit more of the technology stack
15:27
that you're using with open Vino and Intel's
15:30
edge devices. I'm really interested
15:32
to see how some of the smaller guys can
15:34
actually use this sort of technology
15:37
so that it can actually be more competitive.
15:40
Sure, well, leveraging open Veno helps
15:42
us have a real wide range of how
15:45
on the hardware side, how we can install our
15:47
software. What that means for
15:50
smaller manufacturers is that we
15:52
can be quite flexible in the design
15:54
of a system and can accommodate just
15:57
about any budget, which that
15:59
alone is pretty significant to understand.
16:01
I still think there's a misconception that it's
16:03
too expansive or too cumbersome
16:06
for the little guys, so to speak, to really
16:08
innovate in their plants, and it's
16:10
simply not true. You know, we have customers
16:12
that make as little as twenty parts
16:14
of shift, and even
16:17
for them, having the flexibility of how
16:19
we design and configure these systems, it ensures
16:21
that even they can embrace newer technology
16:24
and provide the highest amounts of quality
16:26
to their customers.
16:29
Part of the reason i Can can design and configure
16:31
those systems is because the company uses
16:33
Intel's central processing units
16:36
or CPUs, as opposed to
16:38
GPUs, or graphics processing
16:40
units. GPUs are
16:42
specialized processes are allegedly
16:45
designed to accelerate graphics rendering.
16:47
The key difference in the manufacturing world is
16:50
that CPUs, like the ones Intel
16:52
provides for Iigen, are able
16:54
to perform under harsher or hotter conditions
16:57
like the ones you might find in a factory or factory
17:00
plant. GPUs, meanwhile,
17:02
are prone to overhitting without the use
17:04
of a fan to cool it down, and most
17:06
factories won't use fans so they
17:08
can avoid spreading dust and debris.
17:11
There's always a tradeoff between designing software
17:14
optimized for CPUs or GPUs
17:16
and a manufacturing plant. I
17:18
asked John about this, and I found his answer
17:20
to be quite illuminating.
17:24
It's always an interesting discussion when people ask
17:26
why don't you just go on GPUs and what's
17:29
the real difference? And from
17:31
a manufacturing perspective, just
17:34
logically thinking about what happens in
17:36
a plant. If you remember, like
17:38
late nineties, you remember you had your Compac or
17:40
your Gateway PC, this big old
17:42
white box on the floor, and every
17:44
so often you take the front panel off and it
17:46
would just be totally caked in dust.
17:49
Right, You'd hear the fans humming,
17:51
and well, this is what happens
17:53
to GPUs and factories. This is why
17:55
we don't use fans, because factories
17:58
are dirty. There's dust everywhere.
18:01
And what we found is that when
18:03
we explored using various types
18:05
of mediums, to do our processing. What we found
18:07
is that fanless Intel
18:10
boxes were not only
18:12
just as performant and in some instances probably
18:14
even more beneficial to use. But on
18:17
the maintenance side of it, we
18:19
didn't have to worry about dirt
18:22
and debris, which exists in every single
18:24
plant that we deploy these in. We also
18:26
didn't have to worry about heat. GPUs
18:29
generate tons of heat. I
18:31
had this discussion with somebody who did
18:33
deploy GPUs in a manufacturing environment,
18:35
and they were looking at in tens of millions
18:38
of dollars in HVAC improvements just
18:40
to keep the factories cool enough to
18:42
operate effectively. Right. And then the flexibility,
18:45
like I mentioned, being able to very easily
18:48
scale the hardware for more
18:50
advanced use cases. If we need two or three
18:52
different edge boxes, it's really easy to
18:54
do, and also be able to scale down
18:56
for the smaller applications where we want to make it
18:59
a bit more cost factor for the smaller
19:01
manufacturers as well.
19:04
Coming up next on technically speaking
19:06
and Intel podcast.
19:08
Computer vision specifically for quality is
19:10
becoming more and more common. I think this
19:12
will become completely commonplace
19:14
over the next twelve years.
19:16
We'll be right back after a brief message from our
19:18
partners at Intel.
19:25
Drive away in a new NG this
19:27
end of financial year. The full MG
19:29
range is on offer, including the MGHS
19:32
Essence Suv with a panoramic
19:34
sunroof, sports seats and advanced
19:37
safety tech from just thirty three nine ninety.
19:39
Or get the MG ZST, a
19:42
compact SUV with sporty alloy
19:44
wheels and a ten point one each screen
19:46
from just twenty five four ninety Drive
19:48
away. Speak to your local MG dealer
19:51
to drive away in a new MG this
19:53
end of financial year.
19:58
Welcome back to Technically Speaking, an
20:00
Intel podcast. I'm here now with
20:02
John Weiss. I'd
20:06
actually like to get you to talk a little bit about
20:08
Eigen Innovations. If you could tell
20:10
us a little bit about the company and
20:13
its mission.
20:14
Sure so, Iigen Innovations has been around
20:17
for twelve years. We
20:19
started in academia out of the University
20:21
of New Brunswick. It was founded by a PhD
20:24
student and a professor. We
20:26
started as a system integrator, so we were
20:29
going into factories actually installing vision
20:31
systems, and over the course of about
20:33
a decade we developed our own software
20:36
to make our job as a system integrator
20:38
easier and about I don't
20:40
know. Two and a half years ago or so, we realized
20:42
there's actually a ton of value in IP
20:45
and software we created, and
20:47
so we reinvented the company and moved
20:49
away from leading as a system integrator
20:52
to actually leading as a software SaaS
20:54
based company. We really only do
20:56
one thing. We do inline quality
20:58
inspection, and actually, to more specific,
21:00
our specialty is thermal applications
21:03
that leverage AI. So when
21:05
you think of like injection molding,
21:07
blow molding, metal welding, plastic
21:10
welding, void detection, and
21:12
plastic goods, anything that has
21:14
a heated process that the human eye
21:16
can't easily see defects. We
21:19
do really really well there.
21:21
And we talked a little bit about AI,
21:23
and I think we've also talked about
21:25
the software that utilizes
21:27
machine vision. Where do you see
21:30
AI models and the CPU based
21:32
technology being able to compete
21:35
with machine vision use cases?
21:37
Yeah, it's a good question. Look, I think there are pros
21:39
and cons of both approaches. We
21:42
actually have not yet come across
21:45
a project that we had
21:47
any kind of processing limitation on being
21:50
CPU based. We have applications
21:52
in production running yet thirty
21:55
inferences a second across cameras,
21:57
right, that's quite quite fast.
22:00
There are definitely higher demand applications,
22:03
but in our world a process in discrete manufacturing
22:06
and the types of projects we typically
22:08
focus on, speed has
22:10
actually not been a problem for us
22:12
with CPUs, even at quite aggressive
22:14
speed. I see the tools getting
22:16
easier and easier to use, more
22:19
and more self service, if you will.
22:21
Years ago, we had this phrase of democratizing
22:24
data, if you remember that, around the days of big data,
22:27
kind of empowering everybody to be a data scientist,
22:29
and I see the same movement
22:32
happening in the AI world. In fact,
22:34
actually we're a good example of that. You can use
22:36
our tool to build deploy train models
22:39
across factories and you don't
22:41
have to touch a line of code. So I think that's
22:43
the future. I think the tools get easier and easier
22:45
to use, so that my good
22:47
friend Jimmy who's down in Texas at
22:49
one of our customer plants, who's been in that same
22:52
plant for over thirty years, that
22:55
he can blow me away with how he can build
22:57
a model that does thermal inspection on metal
22:59
wellness, and years ago, somebody
23:02
that didn't have that kind of training from
23:04
a data science perspective or a programming
23:06
perspective, they would never be able to do that, and
23:09
today they're building dashboards
23:11
and building models that are literally
23:13
redefining the way these manufacturers operate.
23:15
It's amazing.
23:18
You heard John say earlier that Eigen has been
23:20
around for more than a decade and this technology
23:23
has been implemented across a variety of manufacturing
23:25
spaces to thermally inspect items
23:28
like metal paper, cardboard,
23:30
box adhesive, automotive windshields,
23:33
and high glass plastics. With
23:35
such a lengthy track record of achievements,
23:38
John spoke about one specific company success
23:40
story that stuck out for him.
23:45
A couple that come to mind. I mentioned
23:47
we inference about thirty images per
23:50
second in this one process. This is a paper
23:52
process, so it's continuous, very
23:54
high speed, and it's for a high
23:56
glass specialty paper. And
23:59
what happens is the high gloss coating
24:01
goes on the paper very rapidly as it's going
24:03
down the line, and unfortunately
24:05
there's a problem where this coding can
24:07
build up and if it's not caught in about
24:09
eight seconds, it will do roughly one hundred and
24:11
twenty thousand dollars worth of damage to the equipment.
24:14
This can happen multiple times as shift. This
24:17
is a very expensive problem if it's
24:19
not caught. And so this one's a great
24:21
example of a thermal application. It's a heated
24:23
coating where we look at that we inference, like
24:25
I mentioned about thirty images a second, and
24:28
in just about one second, we look at
24:30
all of those images, we make a determination is
24:32
there a problem or not, is it good or is it bad?
24:35
And we actually do close loop automation
24:37
as well. We'll send a signal back
24:39
there and trigger a stoppage on the line to avoid
24:41
equipment failure. All of that happens
24:44
in less than one second. So that's a really
24:46
good example of speed. Another good example,
24:48
I'll give you just one more in the interest of time. How
24:51
we can help see things that folks can't
24:53
see. Well, I mentioned fuel tanks,
24:55
and I mentioned some plastic components
24:57
and things like that earlier. Naturally we
24:59
use thermal vision for that humans
25:02
can't see in thermal patterns, of
25:04
course, so we're able to show quality
25:06
engineers inconsistencies in the product
25:08
that they would never be able to see with the human eyes.
25:11
One of our customers manufacturers the
25:13
front plates for a dishwasher
25:16
company. A very large dishwasher manufacturer. And
25:18
so if you've recently gotten a new appliance,
25:21
you probably remember you had to peel all that film
25:23
off, right. Well, what you might not know
25:25
is that film is on from the
25:27
raw material phase and what
25:29
happens is as it goes down the
25:31
process, it gets stamped like a cookie cutter.
25:34
But that film is on it the whole time
25:36
to protect it. So what's really tough
25:38
is for the quality engineers to actually see
25:41
through the blue film or whatever
25:43
tint it might be, to see if there's
25:45
a scratcher dent. And so this is one problem
25:48
we solved for one of our customers where
25:50
they were missing the dents. They were missing the scratches
25:52
because the humans simply couldn't see through the protective
25:54
film. Fast forward to today again,
25:57
another customer that inspects one hundred
25:59
percent of their production on our tooling and
26:01
gives them indicators in real time through
26:04
that blue film if they have any kind of service
26:06
defect.
26:08
And you've talked a little bit about the
26:10
Jenney twelve years ago to
26:12
now, I want to get you to cast
26:14
your mind ahead twelve years in the future.
26:17
Where do you think igen will be and
26:19
in general, where do you think manufacturing
26:22
and quality control technology
26:24
will be in the next twelve years.
26:27
That's a pretty far horizon. I
26:29
don't even know if I could guess the next twelve months,
26:32
to be honest with you, just because the industry
26:34
moves so fast. But let's say over
26:36
the course of the next decade, I would definitely see
26:38
some of the more innovative technologies becoming
26:40
mainstream. So computer vision, there's
26:42
no doubt about it. Computer vision, specifically
26:44
for quality is becoming more and more common.
26:47
I think this will become completely
26:49
commonplace over the next twelve years.
26:51
Often ask this of our guess, but if
26:54
you could have AI solve one thing
26:56
in your field that is manufacturing,
26:58
what would it be.
27:00
I would like to use AI to clone the
27:02
entire Eigen team because these are some
27:04
of the most talented people I've ever worked
27:07
with, and I just need like three to four
27:09
times more of them so I can go take over the world.
27:11
Yeah. Well, we did have an episode
27:14
on digital twins and have
27:16
a human digital twin, so yeah, you never
27:19
know. With that, I'll leave it there.
27:21
Thank you John for your time.
27:23
Well, thank you, this was great. Thanks for having me.
27:27
Thank you to John Weiss for his quality
27:29
insights in today's episode. Of technically
27:31
speaking, in
27:34
a world where we are somewhat preoccupied
27:36
with virtual and digital goods, I
27:39
love hearing stories about the production of real
27:41
world physical products. I
27:43
think we take for granted how much time, effort,
27:45
and brain power it takes not only to
27:47
conceive of new products, but to design
27:49
the whole manufacturing process and
27:52
get them into the hands of you, the customer.
27:55
John highlighted that quality is now non negotiable
27:57
for consumers and that manufacturers
28:00
need to continually reinvest in new technology
28:02
and methods to keep producing high quality
28:04
products as economically as possible.
28:07
A common theme in all of our episodes, and one
28:09
that I'm always exploring, is
28:11
whether these new advances in AI, like
28:14
the machine and computer vision discussed today,
28:16
will help all businesses, regardless
28:18
of size. So it's pleasing to hear John
28:21
say that their technology can help the smaller
28:23
niche manufacturers to use the same
28:25
quality control software and hardware that
28:28
the big players have. This is
28:30
why I'm so bullish about AI and technology
28:32
in general, the ability to lift all
28:34
people and businesses up, no matter
28:37
what stage of life they are in.
28:41
In our next episode, we will look at how we
28:43
can close the AI workforce. Gaut through
28:45
education. So join us on July second
28:48
for the next edition of Technically Speaking and
28:50
Intel podcast. Technically
28:55
Speaking was produced by Ruby Studio from
28:57
iHeartRadio in partnership with Intel
29:00
and hosted by me Graham Class. Our
29:02
Executive producer is Molly Sosher, our
29:05
EP of Post Production is James Foster,
29:08
and our Supervising producer is Nika Swinton.
29:11
This episode was edited by Sierra Spreen
29:14
and written by Nick Firshall.
Podchaser is the ultimate destination for podcast data, search, and discovery. Learn More