Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:01):
Well, we are aiming to become is a platform where
it has a digital twin of your job site, and
our software and our technologies that we're developing is pushing
towards closer and closer to real time data capture. Welcome
to the restless ones. I'm Jonathan Strickland. I've spent more
(00:24):
than a decade really learning about technology, what makes it tick,
and then describing and explaining that to my audience. But
it's the conversations with the world's most unconventional thinkers, the
leaders at the intersection of technology and business, that fascinate
me the most. In partnership with T Mobile for Business,
I explore the unique set of challenges that see I
(00:47):
o S and C t o s face from advancements
in cloud and edge computing, software as a service, Internet
of things, and of course five G. We are often
left wondering how all the leading minds in business continue
to thrive. Let's find out. Our guest today is David Chen,
(01:11):
chief Technology Officer of Skycatch. Skycatch uses drones to capture
precise images of work sites. Like minds, Using drones, the
company can create a three dimensional virtual image of a location.
Skycatch has clients all over the world, and David's job
is to make certain the company's tech can deliver the
(01:32):
solutions to client challenges. David, thank you so much for
joining us today on The Restless Ones. Yeah, thank you
for having me. It's a pleasure. And before we jump
into everything that you do, it's always fun to get
some background. I love to ask this question, what first
got you interested in technology? UM? I think it was
(01:54):
around seventh or eighth grade. My my dad was doing
his p HD at Harbor at the time, so he
had access to dial up through through their They had
a p PP dial up, so this was the real
Internet and I was able to browse the web learn
how to make a web page. Geo Cities was the
(02:15):
first thing I built a web page on. That's why
I learned H T, M, L C, S S and
that's how I got started. Uh. Oh, you're taking me back.
It's like it's like I've got into the literal way
back machine with Geo Cities. What was your first job
in tech? My first job in tech was in high school.
I worked for a local web development company and at
(02:36):
the time, the Internet was still mostly directory base so
Yahoo's and Ulti vis does and um so. The company
I worked for, uh we built our own portal engine,
so that's what I worked on, and I learned PHP
my skill all to kind of build this product and
we actually sold a lot of licenses and I got
(02:57):
a little bit of royalty from each copy that was sold.
So for high school job, that was awesome. Yeah, that's
not a bad first gig at all. I also have
to ask you, David, how did you first get interested
in drones? What was your first experience with those? Um So,
drones has only been I've only been into drones for
(03:17):
the last maybe eight or nine years. However, I've been
into RC aircraft for a lot of my life, and
you know, as a child, I loved airplanes of all kinds,
so it's always been a passion and to learn more
about flight and drones was just amazing when when you
can have something that can just hover in the air
(03:38):
and not crash. So I heard that you were interested
in in drone racing. Yeah, I am still very interested.
So this was kind of how I really got into drones.
Um So, in the very beginning, I got the first
introduction to drones with through this hackathon call Drone Games,
(04:00):
which was actually ran by our CEO, Christian That's how
I met him. Um it was with the one of
the first consumer drones, the Parrot a R. There was
SDK that someone had created that allows you to grab
the camera stream and control the movements. So at the hackathon,
my team built a drone that stream that video to
(04:24):
a computer. It ran some algorithms to detect fist and
tracked it and followed it around. So that was my
first introduction to it, and from there I just got
really deep into drones, started building my own drones, and
the Bay Area has a lot of the top drone
pilots in the world, and there's local meetups all the time,
(04:47):
and I was out drone racing almost every weekend, building
these little racers that can fly upwards of a hundred
miles an hour through these courses. I've seen video footage
of these. I've never actually been able to attend one
of the races myself so far, but the video it's phenomenal.
You the pilots I see who fly these courses, uh,
(05:11):
they see at a speed I am incapable of seeing.
After learning about David's background, I wanted to get a
bit more insight into his work at Skycatch and what
that entails. Skycatch itself is a company built upon emerging technologies,
and so I was really looking forward to hearing David's
(05:32):
perspective on cutting edge tech. So I joined Skycatch in
and when I first joined, I was the second software
engineer we had an intern at the time UM, and
at the time the company was mostly around building the
hardware itself. How do you build a drone that can
autonomously fly to places and take photos and captured the
(05:55):
data that's needed UM, the software was still most sleep
manually processed, so UM I joined to create a platform
for our customers to be able to view the data
on the cloud and to automate that processing. So when
people ask you what your job is, how do you
(06:16):
describe it to them? Today? I think the best way
to describe what I do at skycatches I helped create
the solutions that will solve problems for our customers which
are in mainly in construction and mining. How to save
people time UM, create better safety environments for humans on
(06:36):
the job site UM, and to enable them to make
decisions faster. So so with that in mind, the way
you describe what Skycatch does, how do you put that,
So we build a solution that enables a very high
precision three D capture of your job site. So imagine
a three D snapshot, except every single point on our
(06:59):
on the data that we reduce is accurate to a
few centimeters, and we produce millions of points over any
area that you're you desire to capture, and then you
can then do measurements, to analytics, to all sorts of
calculations on that data after the fact, instead of having
to do that in the field. And I would imagine
(07:19):
that not only saves an enormous amount of time, but
it also uh it ends up being something that really
improves safety as well. Absolutely, a lot of places that
they have to send personnel into for for an open
pit mining operation, there are massive machines, these trucks that
are you know, taller than small buildings. They can't see humans,
(07:40):
so if you have to be near these vehicles, it's
very dangerous, or there's terrain that's inaccessible by foot. Um,
but you can now send the drone up to capture
that data and review that in an office in a
digital form as if you were actually in front of
that area. This is speaking to me because I've talked
to a lot of roboticists who talk about the beauty
(08:03):
of things like robotics and automation is that it helps
take away the dangerous, deadly, and dull tasks that humans
would traditionally have to do and offloads that onto machinery
where you, again, you can do this in a very
efficient and safe way. Yeah. Absolutely. Well. Throughout your career
(08:23):
you've been drawn to, you know, technological companies and entrepreneurial organizations.
So how do you leverage your experience at your your
former workplaces such as Pipio and Twitter in your current
role at Skycatch What sort of things did you learn
that you now apply to your job here. I think
(08:44):
for for my experience at those previous companies, it was
really about honing my technical skills on how to take
a really large and complex problem, break it down into
small pieces, then solve those pieces and put it back
together into a simple solution that makes this complex thing
seem really easy to do. And I think the other
(09:06):
important thing I've learned in my experience is um don't
come up with new solutions, listen to the customers for
what they're real pain points are, and create solutions around that.
And that's what will really make it a successful business.
And you'd be surprised at how many solutions out there
isn't solving a real problem for us. UM. We spend
(09:29):
so much time in the field with our customers. I've
traveled to every corner of the world. I've been up
in the Arctic Circle to visit customers to kind of
understand what is the workflow that they go through when
you're outside in negative thirty degree weather, when there's only
four hours of daylight that you can actually do your work,
and and that kind of helps us drive a solution
(09:51):
that can can make these people happy. Yes, I'm so
glad that it's not the approach of like spinal tap,
where you oh, yeah, this this deal goes to eleven Yeah, exactly. Well,
as a growing company in an emerging field, what sort
of challenges have you tackled as CTO at Skycatch. You've
(10:13):
talked a little bit about this, and but can you
give us sort of a more specific example of a
challenge you faced and sort of the approach that you
and your team went through in order to solve that. Sure,
UM with one of our large customers previous drone mapping, Uh,
everything was less precise because GPS is not very precise.
(10:36):
You can know your position to maybe five to ten
meters in three D space UH, and the data that's
needed is centimeter level accuracy, so they needed these essentially
control markers on the ground to align the data. Are
large customer came to us and told us, we need
you guys to get rid of these markers because it's
(10:59):
taking way more time to set these things up in
the field and then take them down after in order
just to capture data. The drone maybe flies only for
thirty minutes, so we want you guys to get rid
of this part of the workflow. So we spend a
lot of time UM developing custom cameras UH incorporating high
(11:20):
precision GNSS solutions to come with a to come up
with a unified solution that just works out of the
box that you can fly it with no ground control
and achieve that high level of precision on the output data.
And that's was possible only through UM Custom Electronics, custom
(11:40):
software on the computer vision side and making sure all
of this work seamlessly, and control through an iPad app
where the user really just circles an area that they
want hit fly and the thing does everything by itself.
So the goal there is you create an experience for
customer that is almost akin to magic because you have
(12:04):
removed all of the complexities and you've put that on
the back end where you're shouldering that burden. I'm sure
that it took an enormous amount of work on the
back end. Can you give me an idea of of
how long that project took um. I think we started,
we started working with the technologies required in and we
(12:28):
were starting to actually ship the full ready units in.
So for two years we were in heavy R and D,
and for a duration I was flying to Japan almost
on a monthly basis to go testing in the field
with the customer, getting feedback, going back making changes to
hardware software until we got it to a point where
(12:51):
it's works extremely well and extremely reliable. That's incredible. You
went from something that was a very low resolution photograph
to ultra high resolution sution with this particular solution. And
as you say, you know, you're talking about drones that
might be up in the air for less than an hour.
That's not a lot of time to capture that kind
of data. So it is phenomenal to me. That you
(13:13):
were able to do that in a way that not
only gets a full three dimensional picture of the site,
but then can also deliver that data in a way
that's that's consumable to your customer. I mean these are
not these are not easy things to do. Yeah, and
the drone, I think is actually the easier part. The
real magic is in the software. So from the very start,
(13:36):
our CEO, Christian Signs, has always said that we are
a data company, not a drone company. We see drones
as just a tool to capture the data for us.
In the future, it may not be drones to me,
the drone is kind of like this flying tripod that
we can programmatically send to anywhere in space for it
to capture that data. But once that data is captured,
(13:58):
our software does the heavy lifting of turning hundreds or
thousands of photos into these three D structures that represent reality.
That's incredible. If there's one thing most businesses can agree
on these days, it's that change has never come about
(14:20):
so quickly. New ways of working have become the norm.
As a result, the status quo no longer cuts it
when it comes to helping businesses adapt and innovate. That's
why T Mobile for Business uses unconventional thinking to help
businesses work smarter and grow faster. Only T Mobile offers
America's largest and fastest five gene network. It's just one
(14:42):
reason they're better able to help businesses solve the real
world challenges they face as they evolve. For instance, their
new w f X solutions help team members stay connected
and productive where work happens. With nearly two and a
half times the network coverage of A T and T
nearly four times more than Verizon, and forty billion dollars
(15:02):
invested in network and business improvements over the next three years,
T Mobile for Business is better for your business right
now and into the future. See what they can do
for your organization at T mobile dot com. Slash Unconventional
Open Signal awarded to mobile fastest five G network based
on average speeds. USA five G User Experience Report January.
(15:23):
Capable device required coverage not available in some areas. Some
users may require certain planner features see T mobile dot
com m How do you decide upon when to rely on,
say a partner versus developing a tooler platform Internally? Do
(15:44):
you do you do a lot of internal development or
is there a lot of partnership or some combination. UM
this has always been a line that keeps moving. Initially,
for example, when I first joined the company, the hard
problems were around how do you build the drone? How
do you make it fly and automate that UM and
(16:05):
that's where the customer pain points were. This is when
companies were starting to use drones to capture aerial data.
But as drones became easier and easier, we kind of
shifted where that focus was. Because UM as d j
I became a dominant force in the drone industry, we
realized that it doesn't make sense to try to keep
(16:26):
building drones and competing against them. It's better if we
focused on the software pieces, the high precision and work
with them. So that's exactly what we did. We partnered
with them to develop the Explorer one, which is a
custom drone that we designed that sits on top of
one of their airframes with our electronics. Our camera are
(16:48):
g n S S Systems. I see a lot of
businesses that sort of doubled down on their initial decisions
to the detriment of the success of the business over time.
So there are a lot of discussions that went into
the decision to kind of step back from the hardware side. Absolutely,
it was um, you know, it was a very tough
(17:09):
decision for the company because it was so focused on
the hardware side at the time. Um. It was during
a time when regulations around commercial drone used were also
extremely extremely strict, I think around two thousand sixteen, so
a lot of adoption by large entities were not possible
(17:31):
due to these regulations. UM. So we had to think
real hard about what we want to be building, and
ultimately we decided we did not want to be build
building a drone, and we wanted to be building the
software to create value from the data that was captured
by the drone. And and as you point out, because
you did that, it means that the business model you
(17:53):
have is not inherently married to drone technology. It could
be ported over to some other format, some other means
of gathering the data. Yeah, exactly. And we're seeing that
the forms of capture are becoming more and more widespread,
from you know, the robot dogs that can walk a
(18:14):
job site, to embedded cameras on workers helmets, to you know,
even handheld devices. Your your iPhone for example, now has
a light. Our sensors that's capable of capturing three D data.
So all of these sources can be ingested into our
platform for for for visualization and analytics. Well, and that
(18:36):
kind of leads perfectly into my next question, which is
what do you see as the biggest opportunities for Skycatch
in the near term. Well, we are aiming to become
is a platform where it has a digital twin of
your job site, and our software and our technologies that
we're developing is pushing towards closer and closer to real
(18:59):
time data capture. So today you have to fly the
drone capture the data, and then it takes a number
of hours for that data to be processed. We want
to shorten that to real time, and we want to
um be able to continuously capture instead of just having
a single snapshot every day. Well, that's great. This this
is a perfect time for us to to segue over
(19:21):
to emerging technologies because it sounds to me like, well,
we've already started covering a lot of this. It's it
sounds like we're we're talking about things like Internet of things,
we're talking about edge computing. Uh, what are what are
some of the technologies that are just starting to mature
that you're looking toward as being supportive of Skycatch's mission.
(19:44):
So I think two things I'm really excited about. One
is light our technology. UM. These sensors used to cost
tens of thousands of dollars, but now due to the
developments in autonomous vehicles, they've gotten cheaper and small are
I'm looking forward to the days where small camera is
(20:04):
an extremely capable like our capture device that can capture
hundreds of meters at millimeter accuracy UM. The second thing
is UH five G technologies, which will enable the massive
amounts of data that we capture from these devices to
be streamed to edge devices and to our cloud nearly instantaneously. David,
(20:26):
you brought up one of my favorite topics, and that's
five G. I think most people have at least an
idea that five G means high throughput and low latency,
but that's kind of abstract, and the work you're doing
at Skycatch is a fantastic use case for the power
of five G. How do you anticipate leveraging five gs
(20:47):
capabilities at Skycatch? We already have several edge based compute products,
for example, the edge one that's our GNSS based station
plus UM edge compute, so all of the three D
data that's we generate can be processed right on the edge,
so the drone captures it, it's transferred into this small,
(21:08):
small device that just sits on a tripod, and you
can process everything there. However, the bottleneck right now is
once you have this data process, you can use it
in the field, but if you want to share it
with your team, you need to bring this unit back
to the office where you can connect it to the
WiFi or Ethernet and then offload that data. Because a
(21:30):
single flight of a drone can produce upwards of five
ten gigabytes of output data right and having that ability
to transmit data wirelessly changes the game where you can
make those decisions back at home base immediately and then
(21:51):
in theory in the field act upon them. So I
imagine that that would be transformative for many of your clients. Yeah,
I think very much. So we already kind of push
that UM ability to act on the data to the field,
but now being able to connect it to the rest
of the organization is gonna enable so much more to
(22:13):
be done. For example, in in a mining operation, every
time you stop the operation. It's going to cost them
hundreds of thousands of dollars per hour. So and and
they would stop it if something is not correct and
they need to wait to check is this built to
(22:33):
spec is the pit cut to the exact design from
the geotechnical engineers, And if it's not, they need to
go back and rework it. But being able to catch
these issues at the moment that there that is happening
and making that decision right away will definitely save millions
of dollars right and they could even potentially be proactive
and spot something before it has become a problem and
(22:55):
thus avoid causing the issue in the first place. So
this is a really great use case. In what ways
is skycatch leveraging and advancing artificial intelligence? How are you?
How are you? How are you working with AI? So
when I think about AI machine learning, I don't think
(23:16):
about it as this is the solution that's fully autonomous,
it does everything for you. I think of it as
technology that helps humans do things a little bit better
than they're able to do on their own. I'll give
you one example, um when when three D data is
captured from say a mine, a human operator will take
(23:39):
that data into some mining cat software and they may
trace the outlines of the toes and crests of the
high wall. So this is the manual process. They're clicking
point by point until they traced it all. We can
use machine learning to guide the humans. For example, the
human draws a single point at the beginning, and we
can then automatically extend that finding following that ridge or
(24:02):
whatever feature it's following. To take a process that would
normally take an hour to do into something that's just
a single click. And this this sounds to me like
it's the example of when I talk with AI experts,
something they like to talk about is augmented intelligence as
opposed to artificial intelligence. So it's not so much about
(24:23):
creating a system that does things by itself, but rather
lets people do their jobs more effectively. Yeah, and I
think the full potential of AI still yet to be realized.
UM today, the data is still needs to be manually
reviewed and someone is looking at it, deciding what to take,
(24:43):
measurements of what answers are in the in the data itself.
We're working on technologies that will enable to automatically identify
all of the things that you care about, take those measurements,
so when a user is first interacting with at all
the answers are already there. That's that's kind of my
vision for what a I will enable for our customers,
(25:07):
and it's obviously one of those things that has its
own set of challenges. The biggest challenge I see from
from one perspective anyway, with AI and machine learning is
creating a system that is at least somewhat transparent, where
you can understand how the system arrived at whatever solution
(25:27):
it arrived at. Otherwise you run into the challenge of
of something that could look like a black box where
you just have no clue of how the system reached
its conclusion. Therefore, you don't really know if the solution
is valid or if it's applicable. Yeah, you're right, um,
and today, even with solutions that are more autonomous, humans
(25:50):
are still required to really verify that this is indeed
the right answer. Well, while we're on the subject of
challenges and uncertainties, what keeps you up at night? I
think what keeps me up at night is whether or
not we've made the right investment into developing the technologies
(26:10):
that we are Because sometimes these are multi year projects
that we won't see the solution for a long time,
and we want to make sure that this is the
right solution for our customers and for us as a business.
And I would imagine for you that is that is
a big challenge. I mean, we're talking about a a
(26:31):
company that is is itself part of an emerging field
of technologies and applications of technologies. You're doing something new,
so every decision you make is one where you can't
necessarily be certain of what the outcome is going to be.
So I mean, I certainly can empathize with that. Before
(26:53):
I could let David go, of course, I need to
ask him one more thing. Well, then, what do you
think inc is the most misunderstood technology? I think, at
least relevant to me. I think drone deliveries is hugely misunderstood, um,
just because of the complexity of the infrastructure that's actually
(27:16):
needed for this to be widespread will take, I believe
a long time to be really developed and for for
us to be getting our Amazon shipments every day from
a drone. And this is from a guy who I
believe once helped someone by delivering toilet paper to the drone. Yeah,
that was that was a fun joke in the early
(27:38):
days of the pandemic, a roll of toilet paper from
my office route to a friend who lived really close.
I mean, listen, I've been there. I would love to
have a friend who would be able to swoop in
drone with a drone assisted delivery when I need it most.
So I think that it was a valid and important story.
(27:59):
But as as you say, I agree, I think drone
delivery is something that it's far more complex than the
the the individual little use cases that will get Like
I remember when you're in San Francisco. I remember the
story of of of drones being used to deliver tacos.
(28:22):
That was and that was a big, big thing, tacos
from the sky. Um. But I think that that that
sent people who were in the media, like myself down
a road where we were envisioning a future that probably
isn't quite ready to blossom yet. Yeah. I hope to
see this technology truly mature and us to be able
(28:43):
to just order tacos and haven't come in from the sky.
Me too, I mean, I'm always I'm always down for
a sky taco. So bring that future on, is what
I say. David, Thank you so much for joining us today.
It was a pleasure to speak with you. Thank you, Jonathan.
It was really fun. One thing that was clear to
(29:07):
me in my conversation with David was that Skycatches business
is one that will see enormous benefits from five G connectivity.
Being able to deliver a data heavy solution to clients
in real time while conducting aerial surveys in the field
is a powerful message and one that could translate into
significant savings for Skycatches clients. But the really exciting thing
(29:29):
is that it's just one way that five G is
going to transform business in the era of big data.
Being able to transmit information quickly cuts down on response time,
which could mean the difference between heading off a problem
before it can happen and trying to respond to a crisis.
Or it could mean delivering value to customers above and
beyond what they expect, or it leads to solutions behind
(29:52):
the scenes that streamline processes and reduce costs. The possibilities
are endless, and we're going to continue to explore them.
Make sure you tune into the next episode of The
Restless Ones. We're all have more conversations with leaders who
are taking the tech of tomorrow and using it today.
I'm Jonathan Strickland. These days, new ways of working have
(30:20):
become the norm, and the status quo no longer cuts
it when it comes to helping businesses evolve and grow.
That's why T Mobile for Business uses unconventional thinking to
help businesses sees innovation only. T Mobile offers America's largest
and fastest five gene network, which makes their new WFX
solutions possible, letting businesses stay connected and productive where work happens.
(30:44):
See what T Mobile for Business can do for you
at t mobile dot com. Slash Unconventional Open Signal awarded
T mobile fastest five G network based on average speeds.
USA five G User Experience Report, January. Capable device required
coverts not available in some areas. Some users may require
certain planner features see T mobile dock on