November 14, 2024 • 15 mins
Discover how predictive analytics, machine learning, AI, and virtual reality reshape some of the ways we approach design. In this episode, we journey from the origins of predictive analytics to the convergence of big data, IoT, digital twins and more, paving the way for innovative product development. We'll also discuss the potential of virtual reality to enhance collaboration and communication within design processes.
This episode isn't just about embracing the latest tech trends; it's about knowing when simpler solutions will suffice and the critical role of data stewardship. This overview will help you to understand the big picture of where these tools fit into your design process. Listen-in so you can better choose when to use them to optimize your design engineering endeavors, or not.
Join the conversation by sharing your thoughts on our blog or newsletter.
DISCOVER YOUR PRODUCT DEVELOPMENT FOCUS: UNLOCK YOUR IMPACT
Take this quick quiz to cut through the 'design fog' and discover where your greatest potential lies
BI-WEEKLY EPISODES
Subscribe to this show on your favorite provider and Give us a Rating & Review to help others find us!
SELF-PACED COURSE FMEA in Practice: from Plan to Risk-Based Decision Making is enrolling students now. Join over 300 students: Click Here.
MONTHLY DIGEST
Subscribe to the free monthly e-newsletter: newsletter.deeneyenterprises.com
Get the short version on Linked-In: Subscribe here.
About me
Dianna Deeney is a quality advocate for product development with over 25 years of experience in manufacturing. She is president of Deeney Enterprises, LLC, which helps organizations optimize their engineering processes and team perform...
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Dianna Deeney (00:00):
Predictive analytics, machine learning, ai
and virtual reality in designengineering.
Am I really going to try tocover all of this in a 15-20
minute podcast episode?
Yes, I am.
As design engineers, we want touse technology in a smart way
to help us.
However, there's no need toovercomplicate things.
(00:22):
We don't need to use AI togenerate a normal probability
plot, for example.
There are so many options andtools.
Where do we even start todetermine what it is we need?
Sometimes, starting withmethods that have been around
for a while is all that we need.
Other times, we need to pushpast what we've done before and
(00:44):
try a new technique.
The first step is knowingwhat's available and why it
matters to engineering decisions.
So let's talk about all thethings after this brief
introduction.
Hello and welcome to QualityDuring Design, the place to use
quality thinking to createproducts.
Others love for less.
(01:04):
I'm your host, diana Deeney.
I'm a senior-level qualityprofessional and engineer with
over 20 years of experience inmanufacturing and design.
I consult with businesses andcoach individuals on how to
apply quality during design totheir processes.
Listen in and then join us.
Visit qualityduringdesigncom.
(01:28):
Today, we're exploringpredictive analytics, machine
learning, artificialintelligence and virtual reality
, and we're exploring how thesetechnologies intersect and are
used in design engineering.
We had a friend over for dinneron Sunday night and my husband
got out his virtual realityheadset.
(01:48):
Our friend's family wastraveling to Europe and he was
able to use a worldview virtualreality and figure out where it
is.
His family was sightseeing inEurope.
It was like he was hovering ina helicopter, except he was just
hanging out in space andlooking down on the world.
We played around with the VRset with a few more other things
(02:11):
, but the topic came up ofpractical uses for virtual
reality and I was describing tothem how some engineers are
using virtual reality tocommunicate ideas and to make
design decisions.
And that leads to thoughtsabout AI, because everybody's
talking about AI right now.
As excited as I get about thesenew ways of doing things, I
(02:34):
think we need to be smart aboutit.
Just because we can doesn'tmean it's worth doing.
In design engineering, wereally want to use these tools
for many reasons.
One is we want more efficientand effective designs.
We want to optimize our designs.
We want to be able to makedecisions with data-driven
(02:56):
insights.
We want to enhance ourcollaboration and improve our
teamwork.
We want faster productdevelopment cycles.
We want to get our product tothe market in an accelerated way
and, last but not least, wewant to increase our product
quality.
We want to reduce the defectsand improve our product
(03:17):
performance in the field.
With these goals, if weunderstand the unique strengths
of each of these technologiesand if we understand how they
can be combined, we can leveragethese tools to be able to
create those innovative andhigh-quality products that we
want to create.
I want to approach all thesetopics from a development
(03:42):
timeline as far as who was doingwhat, when and how they were
doing it, because they all buildupon each other to sort of a
crescendo at where we are todaywith all of these available
tools.
My background is mechanicalengineering and medical devices,
and I found myself in qualityengineering and doing
(04:05):
reliability engineering workalso.
So the first topic that comesto mind when I think of all
these kind of tools is thepredictive analytics, because
that's what I was using earlieron in my career.
I was using predictiveanalytics.
So what is predictive analytics?
It's analyzing data to be ableto predict what happens next.
(04:30):
It's used in risk assessment,so we're identifying potential
risks in our design andmanufacturing processes and then
deciding to manage them orcontrol them, to manage them or
control them.
Predictive analytics is alsoused for failure analysis, where
we're analyzing historical datato predict potential failures.
(04:51):
This historical data could befield data.
It could be data that wedesigned to collect, in the case
of accelerated life testing,for example.
It can be used for predictivemaintenance and this helps you
determine if there's anyproactive maintenance that you
need to do, like replacingcomponents before they fail.
(05:13):
If you think of statisticalmodeling, that's an early form
of predictive analytics and itrelied on statistical models.
That was the early stages ofpredictive analytics and if
you're looking at a timeline, wecan sort of place that in the
1950s to the 1980s.
Then in the 1990s computersbecame more readily available
(05:37):
and used more often.
So then we got into advancedstatistical techniques like
Weibull analysis.
We started data mining toextract patterns and insights
from large data sets and westarted doing a lot of
simulation and modeling.
Two prime examples of this isfinite element analysis and
(05:59):
computational fluid dynamics.
So predictive analytics iswhere it all started and it's
here to stay, because machinelearning and predictive
analytics are often usedtogether to build predictive
models and forecast futuretrends.
So let's take a closer look atmachine learning Machine
(06:22):
learning.
In our timeline, we could lookat it as starting around the
2010s.
Until now, we're still using it.
When I think of machine learning, I think of those vision
systems that are used onmanufacturing production lines,
where it's used as part ofquality control.
Where a camera is identifyingdefects on a product using image
(06:43):
recognition is identifyingdefects on a product.
Using image recognition, it canbe used in simulations so that
we can optimize the variables inthose simulations.
With machine learning, we arenot explicitly programming
anything.
Instead, we're using algorithmsto allow computers to learn
from data.
It involves training models onlarge data sets to be able to
(07:07):
recognize patterns so that wecan make predictions.
Some of the ways that we dothis training is through
supervised learning,unsupervised learning,
reinforcement learning andneural networks and deep
learning.
One of the things thatnecessitates machine learning is
a very large data set, and thatinvolves big data.
(07:30):
We are living in the big dataera, which involves machine
learning.
It also involves the Internetof Things devices, which are
generating massive amounts ofdata that we can use in machine
learning to provide valuableinsights for predictive
analytics.
Another technique thatgenerates a lot of data is
(07:52):
digital twins.
Digital twins can collectreal-time data from physical
assets, or they can generatesynthetic data through
simulations.
Another source of a lot of datathat you may have heard of is a
digital thread.
That's essentially a digitalrepresentation of a product's
life cycle, from design tomanufacturing to service, and it
(08:17):
connects various stages of theproduct life cycle through a
flow of data.
All of that data can be usedwithin machine learning for
predictive analytics, and thisbrings us to artificial
intelligence.
Machine learning is a subset ofAI, and many AI applications
(08:38):
rely on machine learningtechniques.
However, just because it's anAI doesn't mean that it is using
machine learning techniques.
Here's some of the biggestdifferences AI that uses machine
learning techniques heavilyrelies on data, on large amounts
of data, in order to learnpatterns, and we teach it
(09:02):
through those learning processesthat I just described.
The AI that uses machinelearning techniques can handle
complex tasks like image andspeech recognition, and it can
adapt to new data and improveperformance over time.
The AI that uses non-machinelearning techniques is less
(09:25):
reliant on data and more relianton knowledge bases and rules.
It doesn't learn from data, butit relies on programming and
those rules, so it's lessadaptable.
It requires manual updates tothe rules and the knowledge
bases, and it's right now oftenlimited to simpler tasks or
(09:48):
well-defined problems.
Search algorithms are anon-machine learning technique,
whereas neural networks anddecision trees are a machine
learning technique.
If we continue down the trailof predictive analytics to
machine learning, to AI in thiscase, ai is really helping us to
(10:09):
optimize our activities, tomake it simpler to sort through
data, to get information, tomake decisions.
When we start talking aboutself-driving cars and other
autonomous systems, ai andmachine learning are together
driving the development of thosesystems.
Ai and machine learning aretogether driving the development
of those systems and I thinkthat would require a whole other
(10:30):
podcast episode.
Now, if you remember, at the topof the episode I talked about
having a friend over for dinnerand trying out the VR.
Well, now we've come fullcircle.
The VR kind of started thiswhole thing and now we're going
to end with the virtual realityexperience In product design.
Virtual reality is another modeof communication and another
(10:51):
mode of analysis.
We can help our teammates havethat immersive experience with
our product without being ableto physically tangibly touch it.
Within a 3D world, they couldmove things around and interact
with our product ideas.
This could help us to moreeasily collaborate with our
(11:13):
teammates on the product designsthat we're doing.
It's also another mode ofprototyping, because now,
instead of creating a physicalprototype, we can look at a
virtual prototype.
That may help us reduce costsand accelerate our time to
market.
Some people are using thevirtual reality 3D models to
(11:34):
test product ergonomics.
As of right now, I think youreally have to evaluate whether
or not the immersive 3Dexperience is going to get you
the feedback that you need fromyour customers and your
teammates in order to help youmake design decisions.
Even though it's a virtualmodel and it doesn't cost us in
(11:55):
physical materials, there is alot of other time and resources
invested in creating a 3D model.
How does virtual reality, ai andmachine learning intersect?
Well, ai and machine learningcan be used to enhance the
virtual reality experiences,like generating a realistic
(12:16):
virtual environment or providingintelligent interactions.
So that is a Cliff Notesversion of all these tools for
design.
Engineering, predictiveanalytics, machine learning, ai
and virtual reality are alltools that are mostly available
for us to use, or at least anoption that we might want to
(12:37):
consider when we're doingproduct design.
After this overview of thesetools, what's the insight to
action here?
I think it's to always go backto what it is.
You want to learn.
What is this tool helping youto decide?
In one of the Speaking ofReliability podcast episodes,
(12:58):
fred Schenkelberg told a storyabout an engineer that showed
him that he was excited.
He created an SPC chart usingartificial intelligence and
Fred's response was why?
What's the value in that?
Because with a statisticalprocess control chart, you
really want to map it out realtime to decide if you need to
(13:19):
adjust something, not to collectthe data and then generate a
plot.
The other thing is astatistical process control
chart you can draw by hand.
You don't need to use AI to doit.
With these tools, it's not likelearning a new program language.
If you were of my generation orif you learned coding, you know
(13:42):
that you started to code withsimple tasks and simple outputs
in order to better understandhow to code and how the program
worked.
With these kind of tools, wedon't need to start so simply,
but we do want to have a basicunderstanding of them.
Even though we have machinelearning and AI and virtual
(14:04):
reality, there is still a placefor the basics of predictive
analytics and mocked upprototypes.
But having a basicunderstanding of predictive
analytics and then machinelearning with big data will help
us to understand and decide ifthese new tools are worth it or
(14:25):
if it's overkill, if we need togo back to the simpler
predictive analytics and decidethat that's really all we need.
No matter what, we need to begood stewards of data.
We need to be able tounderstand how to collect, clean
and analyze data so that we canextract valuable insights for
(14:45):
our design decisions.
If you have anything to shareabout these four different
topics predictive analytics,machine learning, ai and virtual
reality leave a comment on thispodcast blog at
qualityduringdesigncom or, ifyou're subscribed to the monthly
newsletter, just respond to theemail from which the newsletter
(15:06):
is sent.
Your message will be sentdirectly to my inbox.
I hope this overview has helpedyou peg and place these
different technologies withinyour world of design engineering
.
This has been a production ofDini Enterprises.
Thanks for listening.
Predictive analytics, machine learning, ai
and virtual reality in designengineering.
Am I really going to try tocover all of this in a 15-20
minute podcast episode?
Yes, I am.
As design engineers, we want touse technology in a smart way
to help us.
However, there's no need toovercomplicate things.
(00:22):
We don't need to use AI togenerate a normal probability
plot, for example.
There are so many options andtools.
Where do we even start todetermine what it is we need?
Sometimes, starting withmethods that have been around
for a while is all that we need.
Other times, we need to pushpast what we've done before and
(00:44):
try a new technique.
The first step is knowingwhat's available and why it
matters to engineering decisions.
So let's talk about all thethings after this brief
introduction.
Hello and welcome to QualityDuring Design, the place to use
quality thinking to createproducts.
Others love for less.
(01:04):
I'm your host, diana Deeney.
I'm a senior-level qualityprofessional and engineer with
over 20 years of experience inmanufacturing and design.
I consult with businesses andcoach individuals on how to
apply quality during design totheir processes.
Listen in and then join us.
Visit qualityduringdesigncom.
(01:28):
Today, we're exploringpredictive analytics, machine
learning, artificialintelligence and virtual reality
, and we're exploring how thesetechnologies intersect and are
used in design engineering.
We had a friend over for dinneron Sunday night and my husband
got out his virtual realityheadset.
(01:48):
Our friend's family wastraveling to Europe and he was
able to use a worldview virtualreality and figure out where it
is.
His family was sightseeing inEurope.
It was like he was hovering ina helicopter, except he was just
hanging out in space andlooking down on the world.
We played around with the VRset with a few more other things
(02:11):
, but the topic came up ofpractical uses for virtual
reality and I was describing tothem how some engineers are
using virtual reality tocommunicate ideas and to make
design decisions.
And that leads to thoughtsabout AI, because everybody's
talking about AI right now.
As excited as I get about thesenew ways of doing things, I
(02:34):
think we need to be smart aboutit.
Just because we can doesn'tmean it's worth doing.
In design engineering, wereally want to use these tools
for many reasons.
One is we want more efficientand effective designs.
We want to optimize our designs.
We want to be able to makedecisions with data-driven
(02:56):
insights.
We want to enhance ourcollaboration and improve our
teamwork.
We want faster productdevelopment cycles.
We want to get our product tothe market in an accelerated way
and, last but not least, wewant to increase our product
quality.
We want to reduce the defectsand improve our product
(03:17):
performance in the field.
With these goals, if weunderstand the unique strengths
of each of these technologiesand if we understand how they
can be combined, we can leveragethese tools to be able to
create those innovative andhigh-quality products that we
want to create.
I want to approach all thesetopics from a development
(03:42):
timeline as far as who was doingwhat, when and how they were
doing it, because they all buildupon each other to sort of a
crescendo at where we are todaywith all of these available
tools.
My background is mechanicalengineering and medical devices,
and I found myself in qualityengineering and doing
(04:05):
reliability engineering workalso.
So the first topic that comesto mind when I think of all
these kind of tools is thepredictive analytics, because
that's what I was using earlieron in my career.
I was using predictiveanalytics.
So what is predictive analytics?
It's analyzing data to be ableto predict what happens next.
(04:30):
It's used in risk assessment,so we're identifying potential
risks in our design andmanufacturing processes and then
deciding to manage them orcontrol them, to manage them or
control them.
Predictive analytics is alsoused for failure analysis, where
we're analyzing historical datato predict potential failures.
(04:51):
This historical data could befield data.
It could be data that wedesigned to collect, in the case
of accelerated life testing,for example.
It can be used for predictivemaintenance and this helps you
determine if there's anyproactive maintenance that you
need to do, like replacingcomponents before they fail.
(05:13):
If you think of statisticalmodeling, that's an early form
of predictive analytics and itrelied on statistical models.
That was the early stages ofpredictive analytics and if
you're looking at a timeline, wecan sort of place that in the
1950s to the 1980s.
Then in the 1990s computersbecame more readily available
(05:37):
and used more often.
So then we got into advancedstatistical techniques like
Weibull analysis.
We started data mining toextract patterns and insights
from large data sets and westarted doing a lot of
simulation and modeling.
Two prime examples of this isfinite element analysis and
(05:59):
computational fluid dynamics.
So predictive analytics iswhere it all started and it's
here to stay, because machinelearning and predictive
analytics are often usedtogether to build predictive
models and forecast futuretrends.
So let's take a closer look atmachine learning Machine
(06:22):
learning.
In our timeline, we could lookat it as starting around the
2010s.
Until now, we're still using it.
When I think of machine learning, I think of those vision
systems that are used onmanufacturing production lines,
where it's used as part ofquality control.
Where a camera is identifyingdefects on a product using image
(06:43):
recognition is identifyingdefects on a product.
Using image recognition, it canbe used in simulations so that
we can optimize the variables inthose simulations.
With machine learning, we arenot explicitly programming
anything.
Instead, we're using algorithmsto allow computers to learn
from data.
It involves training models onlarge data sets to be able to
(07:07):
recognize patterns so that wecan make predictions.
Some of the ways that we dothis training is through
supervised learning,unsupervised learning,
reinforcement learning andneural networks and deep
learning.
One of the things thatnecessitates machine learning is
a very large data set, and thatinvolves big data.
(07:30):
We are living in the big dataera, which involves machine
learning.
It also involves the Internetof Things devices, which are
generating massive amounts ofdata that we can use in machine
learning to provide valuableinsights for predictive
analytics.
Another technique thatgenerates a lot of data is
(07:52):
digital twins.
Digital twins can collectreal-time data from physical
assets, or they can generatesynthetic data through
simulations.
Another source of a lot of datathat you may have heard of is a
digital thread.
That's essentially a digitalrepresentation of a product's
life cycle, from design tomanufacturing to service, and it
(08:17):
connects various stages of theproduct life cycle through a
flow of data.
All of that data can be usedwithin machine learning for
predictive analytics, and thisbrings us to artificial
intelligence.
Machine learning is a subset ofAI, and many AI applications
(08:38):
rely on machine learningtechniques.
However, just because it's anAI doesn't mean that it is using
machine learning techniques.
Here's some of the biggestdifferences AI that uses machine
learning techniques heavilyrelies on data, on large amounts
of data, in order to learnpatterns, and we teach it
(09:02):
through those learning processesthat I just described.
The AI that uses machinelearning techniques can handle
complex tasks like image andspeech recognition, and it can
adapt to new data and improveperformance over time.
The AI that uses non-machinelearning techniques is less
(09:25):
reliant on data and more relianton knowledge bases and rules.
It doesn't learn from data, butit relies on programming and
those rules, so it's lessadaptable.
It requires manual updates tothe rules and the knowledge
bases, and it's right now oftenlimited to simpler tasks or
(09:48):
well-defined problems.
Search algorithms are anon-machine learning technique,
whereas neural networks anddecision trees are a machine
learning technique.
If we continue down the trailof predictive analytics to
machine learning, to AI in thiscase, ai is really helping us to
(10:09):
optimize our activities, tomake it simpler to sort through
data, to get information, tomake decisions.
When we start talking aboutself-driving cars and other
autonomous systems, ai andmachine learning are together
driving the development of thosesystems.
Ai and machine learning aretogether driving the development
of those systems and I thinkthat would require a whole other
(10:30):
podcast episode.
Now, if you remember, at the topof the episode I talked about
having a friend over for dinnerand trying out the VR.
Well, now we've come fullcircle.
The VR kind of started thiswhole thing and now we're going
to end with the virtual realityexperience In product design.
Virtual reality is another modeof communication and another
(10:51):
mode of analysis.
We can help our teammates havethat immersive experience with
our product without being ableto physically tangibly touch it.
Within a 3D world, they couldmove things around and interact
with our product ideas.
This could help us to moreeasily collaborate with our
(11:13):
teammates on the product designsthat we're doing.
It's also another mode ofprototyping, because now,
instead of creating a physicalprototype, we can look at a
virtual prototype.
That may help us reduce costsand accelerate our time to
market.
Some people are using thevirtual reality 3D models to
(11:34):
test product ergonomics.
As of right now, I think youreally have to evaluate whether
or not the immersive 3Dexperience is going to get you
the feedback that you need fromyour customers and your
teammates in order to help youmake design decisions.
Even though it's a virtualmodel and it doesn't cost us in
(11:55):
physical materials, there is alot of other time and resources
invested in creating a 3D model.
How does virtual reality, ai andmachine learning intersect?
Well, ai and machine learningcan be used to enhance the
virtual reality experiences,like generating a realistic
(12:16):
virtual environment or providingintelligent interactions.
So that is a Cliff Notesversion of all these tools for
design.
Engineering, predictiveanalytics, machine learning, ai
and virtual reality are alltools that are mostly available
for us to use, or at least anoption that we might want to
(12:37):
consider when we're doingproduct design.
After this overview of thesetools, what's the insight to
action here?
I think it's to always go backto what it is.
You want to learn.
What is this tool helping youto decide?
In one of the Speaking ofReliability podcast episodes,
(12:58):
fred Schenkelberg told a storyabout an engineer that showed
him that he was excited.
He created an SPC chart usingartificial intelligence and
Fred's response was why?
What's the value in that?
Because with a statisticalprocess control chart, you
really want to map it out realtime to decide if you need to
(13:19):
adjust something, not to collectthe data and then generate a
plot.
The other thing is astatistical process control
chart you can draw by hand.
You don't need to use AI to doit.
With these tools, it's not likelearning a new program language.
If you were of my generation orif you learned coding, you know
(13:42):
that you started to code withsimple tasks and simple outputs
in order to better understandhow to code and how the program
worked.
With these kind of tools, wedon't need to start so simply,
but we do want to have a basicunderstanding of them.
Even though we have machinelearning and AI and virtual
(14:04):
reality, there is still a placefor the basics of predictive
analytics and mocked upprototypes.
But having a basicunderstanding of predictive
analytics and then machinelearning with big data will help
us to understand and decide ifthese new tools are worth it or
(14:25):
if it's overkill, if we need togo back to the simpler
predictive analytics and decidethat that's really all we need.
No matter what, we need to begood stewards of data.
We need to be able tounderstand how to collect, clean
and analyze data so that we canextract valuable insights for
(14:45):
our design decisions.
If you have anything to shareabout these four different
topics predictive analytics,machine learning, ai and virtual
reality leave a comment on thispodcast blog at
qualityduringdesigncom or, ifyou're subscribed to the monthly
newsletter, just respond to theemail from which the newsletter
(15:06):
is sent.
Your message will be sentdirectly to my inbox.
I hope this overview has helpedyou peg and place these
different technologies withinyour world of design engineering
.
This has been a production ofDini Enterprises.
Thanks for listening.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy And Charlamagne Tha God!
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.