Live Talk: Optimizing workflows with Skydio 3D Scan indoor capture
Learn how Skydio 3D Scan indoor capture streamlines digitization workflows, improves measurement accuracy and reduces overall time and costs associated with 3D modeling large factory facilities and indoor assets. Watch this Live Talk replay to:
- Understand Skydio 3D Scan's indoor capture mode, specifically tailored for autonomously scanning large, complex indoor environments.
- Explore how Skydio drones can capture hard-to-reach areas, particularly those that are high up, to generate highly accurate and detailed digital replicas of facilities previously too difficult or complex to attempt.
- Gain insights into the versatile digital twin deliverables that can be created from the Skydio data capture
- Hi. Um, hello, Andre. Good morning, good afternoon, good
- evening, everyone, wherever you're joining on, on today.
- How are you doing, Andre?
- I'm very well, thank you.
- Yeah.
- And good morning, good afternoon, good evening.
- You are based in... You're calling in from Germany, I believe.
- I am.
- So it's evening time for you.
- Yep.
- Morning time here in California. Um, I was actually told that it was always sunny
- in California, but it isn't. It's cloudy today.
- What's the weather like in Germany?
- It's, uh, pretty sunny. It's pretty warm. You should come over.
- [laughs] Thank you. I'll think about it.
- [laughs] Um, so I'm delighted to be doing this
- webinar with you today, um, the expert, the guru, the
- mastermind behind indoor capture.
- I feel, yeah, I feel like I, I'm a massive fan of, uh, of Andre and
- your work. Um, but for those who don't know about you,
- why don't you briefly introduce yourself, Andre?
- Yes. Um, my name is Andre Becker. I'm Product Manager at
- Skydio, and also Head of International Solutions
- Engineering.
- And basically what I do is I develop and implement
- autonomous technology from Skydio into the world and into different industries.
- Mm-hmm.
- Um, and especially when it comes to inspection and mapping.
- Amazing.
- Um, and so
- let's talk about indoor capture, um, and Skydio 3D
- Scan. Like I, I think for those who are not familiar with Skydio,
- um, it might be maybe worth it to briefly
- explain
- what makes Skydio unique in, in the drone space.
- Um, so let's talk a little bit about Skydio
- autonomy.
- Um, maybe you wanna go to
- next slide.
- Yes.
- There we go.
- So a quick refresher in case you haven't seen or heard, uh, about
- Skydio. So Skydio is the US leader, uh, in autonomous drones.
- And, um, as the name just suggests, we believe in
- the autonomy of flying, of flight.
- Um, our drones, besides the main camera, are equipped with six
- additional cameras: three on top, three at the bottom.
- And what you see on the top right is that with these six cameras, we have an
- omnidirectional perception, uh, during flight, so the drone
- sees what's happening around it while flying.
- And we have an AI on board the drone which keeps the
- pilot always safe, uh, while flying, keeps away from
- obstacles, and can also carry out autonomous tasks.
- So one of my jobs, for example, um, is to push our software suite
- 3D Scan,
- where, um, this AI on top of that
- builds, um, automated flight plans for complex
- structures like you see on the bottom right for bridges.
- So you define an area, and the drone will fly
- and make sure to capture every angle
- with multiple pictures.
- Yeah. And this is what is, like, e- like, unique in the drone
- space is, like, we're using, um, incredible technology
- to automate, um, a very extremely
- complex, um, process which is capturing high qua- high
- quality data for inspection-
- Yeah
- ... or mapping purposes. And so the, the drone, Skydio literally does the
- job for you. Um, and, and so for when it
- comes to, like, mapping and inspection purposes, um,
- you don't require any
- previous experience in photogrammetry.
- You don't require any previous experience even in drones, um, or piloting
- drones. So think about-
- Yes
- ... the scalability that that can bring potentially to your,
- to your drone,
- uh, program, um, and your company.
- The way how the technology works, as I mentioned, is, like, highly complex.
- We have an amazing team, and we could talk about, like,
- hours how it
- works.
- [laughs]
- Basically, the easy way to explain it is that you're
- having, like, a professional pilot sitting in the drone
- flying for you.
- So you can be an inspector or a policeman, a
- firefighter. The drone will fly for you.
- So you can direct the drone and say, "Hey, please carry out this job for me,"
- and then the drone will do the rest.
- Just, just to clarify this is we do not have minions in
- Skydio drones. [laughs] This is just, um, very German sense
- of humor here at its best. Um, but yeah, it's a very good way to put it.
- I, I'll have a hard time not to picture
- minions, uh, flying a Skydio
- drone, in a Skydio drone next time that I'm on, on site for sure.
- Yeah. But it does feel that it does the job for you.
- Like, and it does an excellent, an excellent job.
- It does.
- And we'll talk about-
- Yes
- ... more in detail about how the workflow
- works.
- Um,
- um, yeah. So,
- you know, like this indoor capture, so 3D Scan basically is a software,
- uh, enterprise package com- so, it software that comes with the enterprise
- package,
- um, excuse me. And, uh, basically we have different workflows,
- um, for specific use cases. So indoor capture is one of the
- workflows. And I think it highlights, um, something that I
- absolutely adore about, um, you know, working at Skydio.
- Uh, we're on, like, interacting with customers on a daily
- basis,
- and we are very, very good in, you know, ingesting
- and evolving our products ba- based on customer feedback.
- Um, and I think this is a very good example about, um,
- you know, about us.
- So who, you know, what's the industry, what's the use case,
- Andre? What, what was the trigger to develop this indoor
- capture?
- Yeah. Um,
- we consider every customer a partner.
- So we want to make them successful.
- So
- a bridge inspector, for example, is a bridge inspector
- and not a professional pilot
- and if we want to scale drones, we make, have to make technology easier and more
- accessible for them. So 3D Scan provides, for example, a specific scan
- mode to, um, carry out complex inspection of bridges,
- um, and we have that, uh, concept for other industries.
- And in the last year, um, a couple of customers approached us to
- carry
- out this inspection indoors, and-
- The way how 3D scan worked a year ago, uh, it was great outdoors, but
- the, the rules indoors are a little bit different.
- So we have way less space for a drone which moves in 3D
- space. We have less light. Um, we have m-
- Mm-hmm
- ... way more complex, um, structures, um, edges, corners.
- So we really looked at the customer's profile, what they want to do
- with it-
- Mm-hmm
- ... and build indoor capture from it.
- So
- what we can deliver is that if you use indoor capture and 3D scan,
- it will create a digital twin. A lot of customers these days talk about
- digital
- twins. So you can just,
- um, generate it. In this case, it's a warehouse,
- but it could be an old church,
- it could be a factory floor, um, a h- hangar,
- you name it. Um-
- Yeah. And I think this is, uh, you know, the, that space, like the digital
- twin of indoor spaces has been dominated by,
- you know, threshold means, like s- laser scan
- or, you know, expen- more or less expensive equipment, uh,
- but always
- tied to a threshold to the ground. Um, and so this
- one I think is, is just, like opening a whole new world,
- um, you know, think aerial and think autonomous,
- 'cause I think this is the, the
- only way that this is possible is just by, by leveraging, uh, you know, the Skydio
- autonomy, which allows you to safely fly in very
- challenging ex-
- cha- challenging environments such as, you know, a warehouse, um, or,
- you know, facilities of manufacturing sites.
- Um, so yeah, I'm,
- um, really, really, um, e- excited about this and, and
- deep dive into, into the workflow.
- So and that's another, another aspect that we wanted to talk about
- is the ease of
- use of the workflow. Um, so why don't we move
- forwards, uh, Andre?
- Yeah. Let's do it. Just because you mentioned laser scanners,
- I mean, we will not
- replace these traditional methods, um, but indoor capture is
- great where you would otherwise, uh, would use ve-
- more expensive or complex means. So just, like a little example, if you have a
- shelf which is three meters high, like how, like to get the laser scanner
- up there is quite tricky and requires more work, and whereas-
- Mm-hmm
- ... the drone can simply fly there.
- So
- we elevate that inspection for larger indoor environments.
- Um-
- Yeah. And, and that's a very good point, and I, I, I don't, didn't mean at
- any point we are replacing laser scan.
- I think every equipment has a place, um, you know,
- and, and a specific use case, but I think this is presenting just a,
- another tool and a very valid tool to actually get a comprehensive digital twin
- of, of indoor spaces. Yeah, really good point.
- Cool. Let's dive into the workflow.
- I'm seeing we get-
- Yeah
- ... a lot of questions already, and, um, when we go through the workflow, we will
- answer these questions. Um, great questions.
- Done.
- And also at the end we will have enough time to answer these.
- So-
- Yeah
- ... workflow time.
- So a couple of weeks I did, I, I did many indoor captures, but, um, I did
- one with our Skydio 2, um, just in case you haven't seen it.
- Little drone, 770 grams. Um,
- wonderful. And what you will see in a moment
- are screen recordings from
- our controller. So, um, I recorded and I will show you what I saw when
- I was setting up the scan. And I mentioned earlier that we
- want to turn everyone into, like a professional inspector,
- um, carrying out this. So our 3D scan software will always
- carry you through the process of setting up and starting a scan.
- So even if you didn't do it for half a year, that process,
- what you will see on the
- left, will guide you through. So step number one
- is,
- um...
- let's quickly let that give, um, repeat-
- Name your scan [laughs]
- Yeah. So, like the first step, you will see it here,
- we select the scan mode.
- Oh, yeah. Sorry, sorry.
- So depending on the job we want to carry out.
- Yeah.
- Could be a tower capture, could be a 3D capture for complex stuff,
- uh, a mapping
- for a large area-
- And these-
- ...
- or indoor capture.
- Yeah.
- Yeah.
- And these are the different workflows we talked about for specific use cases.
- Um, so if you're-
- Exactly
- ... scanning like a vertical, uh, asset, for example, you have the
- tower, uh, workflow. In this case, indoor capture is one of them.
- Yeah, all belonging to-
- Exactly
- ... 3D scan.
- Yeah.
- Yeah.
- So after we selected the scan mode, we can give the scan a name, and then your
- folder on your SD card will be named like this.
- And, uh,
- after the name we will always add the date and, and the location
- where you scanned.
- Next step is you can set a custom location.
- So when we fly outdoors, usually our GPS antenna handles that.
- Indoors, it works a little bit different.
- So
- almost all the time you don't have a valid GPS location inside
- or underground where our customers also use that.
- So, um, you can type in an address or look it up on the map where
- your scan takes place.
- Then you will see that green, um, that green symbol.
- You s- you can drag your drone around where you
- exactly are in that building.
- You ch- can change the heading, and then we take that position
- and the heading, and
- we write that into all of the images you carry out into the scan into the
- metadata. So later when you reconstruct it,
- you see where that
- happens.
- Yeah. And that's like, uh, basically we're geotagging all of the
- data captures. And so when you import that
- in a photogrammetry software, whichever you process your data, um,
- then your data will be, you know, located at the, at the defined
- location. Um, and it's super-
- Yeah
- ... like I actually love that functionality, that
- if you don't have the location,
- which can happen, you're just doing, um, you know, quick scan somewhere indoors,
- you don't have GPS, um, you can always like look up your, um, your
- address. Um, so that's very, very handy.
- Yeah. The beauty of our drones is we can fly in totally GPS denied
- environments. Uh, we don't suffer from electromagnetic
- interference like by steel
- beams or, um, or wires. So what we do is the drone
- does while it's flying 3D mapping.
- So you will take that location, that will be your origin,
- and let's say you
- fly two meters forward indoors, uh, the drone knows that because
- it does that relocalization in real time, and then we can extrapolate the
- GPS data and it later it will show- Uh, at two meters, he t- also took
- a picture. So you have constant, um, extrapolated GPS
- information, which is great for reconstruction.
- Yeah.
- Next step, we launch the drone. That's pretty simple.
- Yeah, yeah.
- We just, um, press a button.
- Um, once the drone-
- It's done
- ... is in the sky, the drone will ask us for a rally point,
- a return to home point.
- So we can set a position where when the battery is empty, uh, our scan
- is completed
- where the drone will return. Fairly simple.
- I just here use the launch point.
- And let me just add that we have different ways of launching
- the system.
- Um, we can actually hand launch the system by
- just pressing, um,
- three times on the battery. Uh, we can also, we also provide a landing
- pad with Skydio logo, 'cause we have fully autonomous landing as
- well. So that's, you can do a case or, um, the landing pad.
- And so you can also place the drone there and, uh, launch from
- whatever. So it's, it's also, like, really interesting because in an indoor
- space,
- uh, in an indoor environment, excuse me, the space that you have for a landing
- and takeoff might be quite constrained.
- Um, and thanks to not only the size of the,
- uh, of the Skydio, but also
- the autonomy, um, we can safely do so in very constrained
- spaces.
- Exactly.
- Yeah.
- Cool.
- Next step. So before a s- a scan starts, we have to
- define a volume. So we have to basically tell the drone,
- "Listen, this is the area I want to cover with
- enough
- pictures to create a digital twin." And the way how we do this
- is to define three things. A floor, like the lowest point in our
- volume, the highest floor, and that we set a fence.
- So we start with the floor. Um, so I fly to the lowest, lowest
- point. I will allow the drone to fly here.
- Then go to the next step, set the ceiling.
- It tells me fly to the top of your desired scan volume.
- So I fly under the ceiling.
- Since the obstacle avoidance is always activated,
- the drone will simply stop,
- uh,
- 1.2 meters, um, in front of the roof, and I can set my ceiling.
- Mm-hmm. Um, and this is like, um,
- like one of the best uses that I've seen of mixed reality.
- Like you're interacting, like Andrea showed before, like you're,
- you're using the
- controller, so you're interacting with, you know, the drone, what the drone is
- seeing, um, and also using that mixed reality to actually help the,
- the
- workflow be super intuitive and super easy to, to
- follow. So, um, so yeah.
- You define what to be scanned. So in this case, in this ex-
- Mm-hmm
- ... example, I wanted to, um, um, scan the entire room.
- But let's say you have a big factory hall and a machine which is 20 meters high.
- I'm making this up. So you can set the floor at
- 10 meters and the ceiling at 25 meters.
- So you will only then scan the upper area of the machine.
- So it's your choice. Scan the entire room
- or just parts of an indoor
- environment.
- Yeah.
- And this is how we tell the drone what to scan.
- And the ceiling, uh, just to emphasize, the ceiling will also, uh, it
- can become a geo fence. And in indoor spaces this is super
- important, 'cause typically you'll have, um, you know, a lot of like thin
- wires or structures that, um, it might be just the risk of,
- of collision, it increases. Um, and so, uh,
- that, that ceiling can, can w- you can create like a geo fence, and
- Andrea will explain this in, in-
- Yeah
- ... in, uh, in, uh-
- Yeah
- ... in the next steps. Yeah.
- That's a very good point. So a lot of times indoors you have these
- very
- small wires hanging down, so normal power cables shouldn't
- be a problem. But, like, these thinner wires are a problem.
- Our drone navigates visually, so with cameras.
- So everything our eyes can barely see, the drone also will barely see
- and potentially fly into that, and we want to avoid this.
- So my recommendation, if you have these lights,
- just fly a little bit under them
- and set the ceiling under these wires, under these lights. Later,
- we'll set it up.
- The drone will still look up and take pictures of the roof,
- but this ceiling really
- is also a limit for the drone and
- to not fly above that.
- Exactly. What I typically do is actually flying, um, to the location
- that I feel comfortable with, and then set the ceiling at that,
- at that height. Um, and then obviously, uh, set up the geo fence for
- it. Um, but yeah, I can send... I've, I've been seeing a few questions
- about the obstacle avoidance. I can send you.
- There is, um, a blog post, uh, with the different settings.
- So we have three settings for the obstacle avoidance.
- Obs- obstacle avoidance is, um, active by default at all
- times. Um, but we do have different settings in case
- the drone
- is navigating, um, in, in more confined spaces.
- So it creates basically a bubble, um, a 360 bubble,
- and it just, it will avoid any obstacle, um, for
- that, th- th- those three specific settings,
- and it comes, the bubble basically
- comes closer.
- Exactly.
- Mm-hmm.
- Next step,
- we set the boundaries of our volume.
- So imagine you've just bo- bought a new property and you want to,
- like, limit the
- property, so you go around your property
- and set a fence.
- It's exactly how it works here. So you fly around the room,
- and then you can set these pillars with that blue button, and you can
- define a volume like you want. So in this case, again, I
- want to capture the entire warehouse, so I fly to the edges and set
- these pillars.
- Um, it's also a really cool example how, uh, we overlay this in AR.
- So since the drone does 3D mapping, so we can really place that
- pillar in 3D space. You can really fly precisely.
- Yeah, and the pillars basically follow you.
- Um, and all the mixed reality you are seeing,
- so all the elements you're seeing
- on screen, they're also like the, the screen is obviously a touch screen.
- Uh, so you can drag, um, you know, the ceiling, the floor, and even the
- pillars on the screen to refine the position.
- Um, so yeah, it's, it's pretty cool application. Yeah.
- Yeah. And basically with these three steps,
- the floor, the ceiling, and these
- boundaries, we defined a volume, like an inside volume,
- um, and told the drone-
- ... like what to focus on.
- Yeah.
- In the next steps, we can make that volume bigger.
- So obviously we want to capture the entire warehouse, and we cannot
- fly behind the walls to include that in the volume, so therefore
- we, we, um,
- we separate everything which is inside these boundaries and what's
- outside these boundaries. And with this setting you can say,
- um, please also activate everything which is above the ceiling,
- this would be our roof, or below the floor,
- this would be the actual floor,
- and outside of the wall. So if we activate these
- outside pillars then the volume gets larger.
- Um,
- I made a little drawing just to make this super clear.
- 'Cause... Yeah, 'cause this is a bit confusing.
- It was a bit confusing for me at the beginning, so, [laughs]
- It is
- ... and, and Andre is very good at drawings.
- [laughs]
- It is. Um, it is a little bit complex to understand,
- but
- we really try to make indoor capture as versatile as possible.
- Mm-hmm.
- And, um, once I explain that I t- I tell you, like, when to use
- what. So this is a room from above, um,
- just 2D. So what I do, what we just did, we fly to the
- corners and set these pillars-
- Mm-hmm
- ... then we have this shape. So everything which
- is inside the blue rectangle is
- our inside volume, and by default the drone will scan
- everything which
- is inside. So could be machines, could be racks, you name it.
- With this setting you see on the left above ceiling, outside walls, below floor,
- we basically take this blue area and virtually push it behind
- the walls and then all of a sudden
- it, like, the walls and everything which is outside is included.
- And then basically the drone will scan everything it sees in that,
- in these boundaries while always flying inside the
- blue area to, to keep safe. We'll come to that in the next step.
- Yeah. So even if you can... So, so basically, let me see.
- Um, if even if you don't place the boundaries
- as, you know,
- the extreme locations on the corners, for example, of your warehouse, this
- allows you to ensure that all of the data, the corners,
- um, are gonna be captured and covers in the flights,
- um, in the, in the, in the mission, basically.
- Yeah, pretty much.
- Yeah. Okay.
- Two different job profiles. So-
- Mm-hmm
- ... job number one is my job. I want to create a digital twin of the
- entire indoor environment.
- Mm-hmm.
- So, um, I do it exactly like I just explained.
- I set the boundaries at the edge of the room,
- um, and I expand it, and I have everything inside and outside.
- Let's say I'm an, I'm a inspection crew
- and I only care about
- the facility. Like, is the roof in good shape?
- Are the walls in good shape?
- I don't care, um, about what's inside the hall,
- like the machines and stuff,
- only what's outside. In this case, I would just turn off
- the blue rectangle, and the drone would then only focus on the
- outside blue to just capture
- the room or, like, the hall, basically.
- This is what I meant, um,
- when I said we wanted to make it, uh, as versatile as possible.
- Yeah. True.
- These are the two concepts here.
- Um, I have a couple of questions. Um, you know, there's one
- about light conditions. Um, maybe we can talk a little bit about
- that, um, because obviously you mentioned, Andre, already
- that, uh, we navigate, uh, based on the six navigation cameras.
- So our navigation is b- is purely based on visual, uh,
- information.
- So, uh, the light conditions, although I've, I've, I've flown in, in,
- you
- know, warehouses and, and indoors facilities,
- light conditions are challenging, um,
- but obviously you do need, um, you know, a certain amount of light to be
- able to navigate and also capture the data.
- 'Cause if you don't have good enough light conditions, the data you're gonna be
- capturing, um, isn't, like, is not gonna be...
- You're not gonna be able to process that and generate a 3D model.
- Yes.
- Um,
- it's a good question, light conditions.
- So-
- Mm
- ... the drone navigates visually.
- I mean, the drone has a GPS antenna, we can't use it indoors, and
- also other sensors, but mainly we use cameras.
- Cameras need light. Um, if it's totally dark, you simply can't
- fly. The... But even in fairly darker
- environments or darker warehouses, you can fly.
- And if it's getting too dark, the drone will tell you.
- So if it's too dark, for example, the drone will not take off.
- It will tell you, "Please move to a, a more brighter area."
- Mm-hmm.
- And also if you fly, um, across your indoor environment and it,
- there's a section where it's too dark, the drone will tell you
- it's too dark before it flies into that area.
- Um, so you're always,
- you're always safe.
- Yeah.
- Um-
- And just to make a, like, um, so you, you could still fly the
- Skydio as a dumb drone, um, but you will not have the, uh,
- obviously the Skydio Autonomy. So that, that's, that's where it's,
- it doesn't work.
- Exactly.
- Um,
- but also from a data perspective, like, it doesn't make any sense.
- If you're capturing data, um, you do, you do need, uh, light
- conditions, like good enough light conditions.
- Exactly. Yeah. Um, but you would be surprised in how dark
- environments it, it works.
- Yeah. [laughs]
- Um, it also surprised me, and it's also part of the reason why it
- took us around a
- year to develop it because we optimized also the exposure algorithms to
- always take a picture which is
- well exposed, uh, and while the drone moves, uh-
- Mm-hmm
- ... also without motion blur.
- It's not as easy as it, as it sounds, but once the drone flies, it will
- take the perfect picture for you even if it's dark.
- So you have to-
- Yeah
- ... you don't have to worry about that.
- Yeah.
- Cool.
- Cool. So what's next step?
- Yes. So, um, you already mentioned it, a
- geofence.
- Mm-hmm.
- So by default the drone will not the, will not leave the
- area you set up. So the drone will always stay inside these
- boundaries you set up and always under the ceiling and always
- above the floor.
- Let's say you, you scan a parking lot, you have large openings,
- or a factory with a
- large opening,
- we don't want the drone to just exit and scan, like, the
- neighboring room.
- Uh, or like fly out of a parking lot and scan from outside.
- We want to stay inside, and tha- therefore
- this geo fence is turned on by default, and with that volume you set up, you can
- control where the drone flies.
- Yeah.
- Um, if it's-
- So basically you're saying you, you're just asking the drone to stay in
- the
- contain, in the contained area, in the contained space.
- Um, so if you had, to follow your example, Andre, if you had like a, a
- parking space or even like a, like a, a, a door or
- a gate into a different room and you don't, uh, have
- a geo fence and you have extended area, the drone will, will, will
- fly and try to capture the extended area as requested.
- Um, but I, I actually, I think that this is very good that it's always
- on by default, um, just 'cause as we mentioned, the indoor
- spaces are normally challenging in terms of, um, you know, obstacles
- and whatnot.
- Exactly, yeah. So, um, indoor environments
- can propose some
- more dangers, like moving people. We don't wanna hurt them.
- We don't want the, the drone to exit a volume,
- so that's the reason why we
- have this in place. So every operator controls where the
- drone is allowed to fly and where not.
- You can turn this off, um, if you want t- uh, if you have a super controlled
- environment, and then the drone will fly also outside
- that,
- uh, volume to capture more angles if you want that.
- Yeah.
- Cool. Now comes the really fun part.
- So now we set up everything we need to kind of start the scan.
- Um, the next step is the exploration.
- So we hit a button, and then from there the drone flies by
- itself.
- It's always the coolest part because the drone will fly inside of that
- volume now, and it will explore the volume.
- So every structure is different, and, um, you see it here.
- The drone will build a 3D mo- model, like a low-poly mesh
- while it's flying, um,
- before the scan even starts.
- And this happens very quickly. This ha- this happens in matter of minutes,
- depending on how big the, um, the space is, but it might take, like,
- less than a minutes. Um, and as Andre,
- uh, highlighted, it basically just, um, wants to
- understand the geometry of the space, um, so then
- we can define the settings for the actual data capture.
- So right now it's not capturing data.
- It's just understanding the, um, geometry-
- Yeah
- ... of the assets that we've, uh, required to inspect or we've defined
- to inspect.
- Yeah. We always have to think about the outcome,
- uh, more downstream.
- So in order to reconstruct it in photogrammetry, we always
- need kind of the same distance to our object, and we always need the images to
- have, like, a certain overlap and also side lap.
- And if we can just, if we would just plan a regular flight
- plan based
- on your volume, we don't know, are there objects, um, in that volume?
- And we would produce gaps, and that's the reason why the drone flies.
- It scans the geometry, and based on that geometry, it creates
- that tailor-made flight plan for you, so you can make sure that you
- have covered every angle possible.
- Yeah. And, and this is, this is what is, what makes 3D scan
- absolutely unique, um, and also enables that
- scalability that we talked about earlier because you
- re- like, the skills that you need to actually capture this
- manually
- are... Like, I couldn't do it. Like, as in absolutely no, no questions asked.
- Like, you, you will require a massive amount of
- training to capture the right
- data and, you know, to then generate a comprehensive 3D
- model/digital-
- Yeah
- ... digital twin. Yeah.
- I mean, doing photogrammetry of drones is not a new process, but it's usually-
- No
- ... very cumbersome and also error-prone.
- And if you have to, um, shoot thou- thousands of pictures with the same
- distance
- and overlap, it's quite tiring. And especially indoors,
- most of the people are busy with keeping the drone away from
- obstacles with regular
- drones because indoors, uh, they fly a little unstable, let's
- say. Um, and then all of a sudden, after the flight
- in your office, in photogrammetry, you, you see,
- oh, I forgot this corner, for example.
- This will not happen with 3D scan because it's a machine.
- Uh, the AI will plan
- a gapless scan for you.
- Yeah. And, and we'll see now, if you go to next step during the data
- capture-
- Yeah
- ... we'll talk about different couple of tools
- that we have-
- Yeah
- ... in place. So in case you need to capture
- extra data, you can, you can do so.
- Yeah. So the drone just came back from the
- exploration.
- It knows now how big the volume is, uh, what kind of structures we have inside,
- and now we can set up our scan. So we can set up the distance to the
- objects, the overlap, and the side lap. So in the middle you see it.
- I see before I even start the scan what will be my GSD,
- how many photos will the drone take for me, how long will it take, how many
- batteries will I need. So I can adjust these parameters before I
- even start.
- Yeah. And, and just a quick note on this,
- the overlap, obviously both overlap and GSD, the ground sample
- distance, are gonna have a massive influence in
- terms of, uh, the time
- and the batteries you're gonna be requiring to capture your size.
- Um, but please be mindful of the settings
- that you
- need, uh, the overlap settings that you need to,
- the overlap requirements, excuse
- me, that you need for your photogrammetry
- platform.
- 'Cause that we've seen, uh, we've tested, uh, you know, the main
- photogrammetry plat- platforms like Drone Deploy, Pix4D,
- ContextCapture,
- RealityCapture, and the requirements differ from one to
- an- to another one.
- Um, so going from 70% to 80% overlap has a
- massive
- influence, um, but you do need to do so depending on the
- photogrammetry platform that you are using.
- Yeah, exactly. So
- I also, uh, had success with 60/60 overlap in a, a
- certain, uh, photogrammetry solution.
- I will not name-
- [laughs]
- ... uh, the partner, but-
- I know
- ...
- uh, yeah. And-
- It's incredible actually. There has been massive improvements in
- that space as
- well, and there is a few software packages
- that are, uh, providing
- extremely good results for, for indoor, um, for
- indoor data-
- Yeah
- ... which is, yeah, very, very good to see.
- Cool.
- So I just hit begin scan once I-
- Yeah
- ... set up these settings, and then,
- um, the drone will start capturing. So it's hands-off from that moment.
- Uh,
- and the drone will fly by itself. You see, um, the pictures it will take along
- that, and it will fly along your indoor
- structure.
- And you see like the gimbal is always tilted to capture the surface
- with the right angle.
- And
- yeah, so
- this is-
- Yeah
- ... the, the beauty of 3D scan.
- Are you, are you-
- So every, every, um-
- Alicia is going- Excuse me just give me one sec.
- Yeah. So while Alicia is figuring out what happen-
- what happens behind her camera,
- the drone continues at that mission.
- So, um,
- it doesn't matter if it's the wall, the floor, the ceiling, or racks,
- the drone
- will fly along and capture these images, always keeping the same distance,
- uh, always the same overlap and side lap.
- And again, we overlay these waypoints where the drone
- will fly next, uh,
- onto the screen while flying. But
- again, you don't have to do anything at this
- process.
- Um, the drone does it for you, all fully automated.
- And once your scan is completed, um,
- we cut to this. The drone comes back
- autonomously, uh, where you started, where you set up the rally point,
- um, at the beginning of your scan. You still have the chance to fly
- manually and add some pictures before you land, but this is
- basically the entire workflow. And, um, Alicia also mentioned, um,
- our landing pad. So we recognize that specific Skydio logo you see
- here in the video. So once the drone sees that,
- it will always land,
- um, on this pad, and
- then your indoor capture is complete.
- So the host has left me. No. Has the host left me?
- [laughs]
- Sorry about that. I'm in a, I'm in a conference room, and, um,
- someone was just connecting to this conference room, so I
- was hearing voices in the
- background. So apologies for that.
- [laughs]
- No problem. You will be, uh, happy to hear that we, uh,
- finished our indoor
- capture.
- [laughs] Amazing.
- Yeah.
- Apologies for that.
- So basically what happens after the scan is
- you have, um, a lot of pictures now onto your SD card,
- and you take these images. Um, I recommend Skydio, uh,
- Media Sync, so where, um, the drone are synced automatically to the cloud,
- and we
- have some partners where we automatically give the data to.
- And from there, in any third party, um, photogrammetry solution, like
- the biggest in the market, we support them all.
- And you can create your digital replica.
- Mm-hmm.
- In, in preparation of this webinar, I did that.
- So, um, as I said, I mean, we work with many
- solutions, um, Bentley ContextCapture,
- RealityCapture, Trendspek, um, a lot
- of partners.
- DroneDeploy, Pix4D, GenX.
- Exactly.
- Um, we like-- I, I think we're, we're completely agnostic, um,
- to the photogrammetry platform, uh, our customers. So it's entirely up to you.
- We help you navigate that ecosystem as well because every
- platform there is
- no, um, you know, one size fits all.
- So we, we help you choose depending on your
- requirements and your workflows.
- Um, but yeah, this one here,
- uh, Andre, you wanna, you wanna talk a little bit how you
- generated this?
- Yeah. Um,
- so, um,
- it depends always on the photogrammetry solution-
- Mm-hmm
- ... um, wh- which you use. Um, in this case, what I show you here is
- DroneDeploy, um, where I simply upload the data.
- We also have a, an integration. When I uploaded the data from Skydio
- Cloud,
- um,
- I, um, I can just start my model build.
- If you don't use Media Sync, um, take the images, upload it.
- Like it would work with like WeTransfer style.
- Um,
- and then DroneDeploy will send you an email once your model is ready.
- So you don't need to be a pho- a professional photogrammetrist,
- and the model will
- just appear.
- We actually have a, um, an API integration with DroneDeploy, so
- actually there is no human interaction whatsoever required for the process of
- your
- three M model once the data is uploaded into the cloud, um,
- which is, uh, you know, highly
- valuable if you're, if you're like...
- By, by the time you go back to the office, uh, you could have your three D model
- ready to, to download and, um, you know, consume
- in DroneDeploy. And we also have, um, you know, other
- APIs and, and integrations coming up.
- So stay, stay tuned for that. Um, Andre,
- can we, um, can I go over a few questions that
- I'm seeing coming through-
- Yes
- ... before-- Yeah.
- Yes.
- Um,
- we have a, um, a couple of other questions about light
- conditions. I think we covered those quite well.
- Um, I've seen comments about tunnels and mining applications.
- So, um, yeah, it's, it's always gonna be
- if you don't have good
- enough, uh, light conditions, uh, you won't be able to use Skydio
- Autonomy, um, and also you won't be able to capture,
- you know,
- quality data for your three D modeling, uh, process.
- Um,
- there was a question about overlap.
- Is there a point where the over and side lap amount creates just more
- images
- than-- that aren't useful to the end mesh?
- Um, that's not related to Skydio, I will say.
- Um, Andre, and I will, I will let you comment on this, but I,
- I did
- a few tests in the past in terms of like the influence of the, of the overlap
- in the quality in the, in the three D model,
- the final quality of the three D
- model, um, and there is a point where it doesn't increase.
- Um, back in the day, we had to use, you know, way higher overlap.
- These days we're seeing, as Andre mentioned,
- we're seeing very good results with
- even sixty percent, um, sideward and forward overlap.
- Yeah. Um, it's a very good question.
- Um-
- If you have time, I recommend always go through the images
- if
- you have the time just to see if everything-
- Mm
- ... is good enough. I mean, our system works pretty good,
- but from 1,000
- images, one or two could be blurry.
- Usually a photogrammetry solution will just throw this, uh, image out,
- and you have
- enough other images to, um, to cover in that place.
- Um,
- and
- to answer the question, like, personally with 3D scan, I've not
- seen a reconstruction, um, across the platforms
- where too many pictures pose the problem.
- The-
- Yeah
- ... the reason for that is that
- since the drone knows exactly where it was, it will always write a
- perfect location into the metadata, and that will help greatly during
- photogrammetry. So, um,
- it's working pretty good.
- Yeah. Um, we also provide, like, after the flight you get the pictures,
- then you get the capture reports, and then you get a glTF file
- that you can drop into gltf.com, something like that,
- a viewer, online viewer, and that gives you also as well a kind of
- like a, um, a coverage map, um, so you can
- h- you can have a look if there's a, any areas that you need to capture
- additional
- pictures or videos.
- Um,
- okay, so let's move on, um, to another
- question.
- 3D model. Oh, question.
- Oh, 3D model. Oh, you wanna... No, l- let's go through the 3D model and,
- and I'll
- go through the questions in the meantime.
- Cool.
- Is that okay?
- I mean, so yeah, sure. So in this case, um, Drone Deploy built me, um,
- a great model, um, which I can access in the browser.
- Um-
- Mm-hmm
- ... the cool thing, I have all the inspection tools, um, in,
- built in
- here already. So, um,
- from this model, I can do a lot of different things.
- So, uh, first of all, I can...
- You see I put already a couple of annotations here in the model.
- So, um, let's say
- I can... Here you see all the camera frustum,
- so the, the position where the drone
- was. So I can just click to a position like
- here on the
- top of the shelf, and then Drone Deploy will show me an
- image
- how it looked. And then I can also, like-
- Mm
- ... go sideways and look at different aspects.
- So I always see the original picture the drone took.
- Um-
- Mm-hmm
- ... and in this case I was looking for a specific cable, which
- I found. I mean, it's just an example case.
- Um-
- Yeah
- ... I can just-
- Yeah
- ... mark, for example, issues or observations.
- So I'm also looking for this cable in, in this instance and I can say, "Hey, what
- is it? Is it a missing part? Is it rust?
- Is it a safety hazard?" So I can just-
- Yeah
- ... classify.
- In this case, I say, "It's incomplete work. Why
- is the cable here?
- Uh, remove it."
- Uh-
- And that's,
- it's, it's, uh-
- Yeah
- ... e- extremely useful actually to, to have the 3D model linked to the images
- because, um-
- Yes
- ... as you saw earlier, the GSD is gonna be somewhere in the millimeter
- range. Um, but the 3D model
- resolution will sometimes not allow you to see labels, for
- example. Um, so-
- Yeah
- ... but if you click on the 3D model, then the pictures come up,
- and then you are,
- uh, you can, you can, you can see all the labels and any,
- you know, smaller
- numbers. So-
- Yeah
- ... um, that's a very good application-
- Yeah
- ... for, for this.
- This is a good example here. I mean, this is just-
- Mm
- ... totally made up, but the concrete.
- So, um, this is under the roof, and if you just take a normal picture,
- which is not
- a reference in 3D space, this could be anywhere in a large factory.
- You don't know where. But these various inspection tools allow
- you to
- mark these damages, uh, where you can report later on.
- But you not only have this in the image now, but I see exactly where is it.
- So, um-
- Yeah
- ... if a repair crew has this... Whoa. Let me move out of that.
- If a repair crew,
- um, has the model, they can see exactly where the
- damage is.
- And-
- Yeah. Th- th- that's a bit of a-
- Speaking of reports. Yeah.
- Mm.
- I can just everything I marked, for example-
- Mm
- ... here, um, I, this is a classic inspection of safety
- equipment.
- I can just mark these things, and I can find them later.
- Yeah. Yeah, and this, that's a good example of, like, you know,
- reading small,
- small labels and, and letters. Um, what are the formats output we
- can expect? So we're talking... So Skydio provides the capture
- data, so not the photogrammetry. We don't offer the photogrammetry,
- uh, model. So you will have to take those images.
- Um, so we provide capture reports.
- Um, we provide a, uh, that glTF file to,
- for the coverage m- the coverage map, um, and then the pictures in
- JPEG formats, and then we have a low resolution video,
- um,
- as well for every flight we, um, we do.
- Um-
- Exactly
- ... we have-
- Yeah.
- Yeah, sorry, Andre. Go for it.
- Our product takes care of that you can generate a
- perfect data set.
- Mm-hmm.
- Without the perfect data, without the, the right overlap and side lap,
- photogrammetry reconstruction is impossible. So this is our job.
- We deliver these pictures, and once we have a model,
- um, you can export various file formats.
- So in Drone Deploy alone,
- you see what I can, um, export here.
- I can, um, export 2D maps, 3D models as
- OBJs,
- um, point clouds,
- you name it. And all the other players, the big players here, they c- they deliver
- the same outcome once you reconstructed the model.
- Yeah.
- Um, yes, the metadata is, um, tagged into
- every image. So I'm not sure, Adam, why you experienced any issues, but,
- um, I'm pretty sure, and I've been using ContextCapture, uh,
- for a long time now,
- you should not have any issues and metadata should be, uh, there.
- I think Pix4D is the only one that you need to import an additional
- file,
- which we provide with every flight that we do.
- Um, we have, um... Let me
- see.
- Uh, GenX has the cloud integration too.
- Yes, uh-
- Yeah
- ... Ross, thank you so much for this comment.
- Um, we do have a cloud integration, API integration with, with GenX as
- well.
- Um-
- And t- t- ch, let me see. Another few questions more a-
- about... Well, Aiden was replying to, um, you know, the minimum
- amount, amount of lights, um, and the behavior of the, uh,
- the drone in low-light environments.
- But do remember that at the end of the day you're capturing data, so in
- very low light conditions even if the drone can fly,
- the data you're gonna be capturing is not gonna be good enough for you to
- read labels, to do inspections, or to do 3D modeling.
- Yeah. Um, it's a good point. And as I mentioned before,
- the drone will tell you
- when it's getting too dark. Um-
- Mm-hmm
- ... in the end you want to, um, generate good data as well.
- Um, so
- let's talk about the obvious. So Indoor Capture is a fantastic tool for
- indoor
- environments, but, um, I personally love
- drones, but I also know the drone is not the solution for every application.
- So
- just to make it really clear where Indoor Capture is maybe not the best solution
- you could use here is in small environments.
- So like-
- Mm-hmm
- ... a living room, like small residen- residential rooms and stuff.
- To capture this, the drone simply does not have enough
- space
- to safely fly and maneuver. Um,
- don't.
- Yeah.
- I wouldn't do it. Um-
- Yeah. And, uh, and I think we, we, like, w- we make emphasis
- on this. Like, we don't want you to use Skydio in,
- you know, p- uh,
- like, they, there is a tool for, for, for every use case.
- And so be also mindful that, you know, um, very small indoor
- environments, this is, you know, don't, don't, don't, like, try to do a
- 3D scan of your, you know, of your house, of your apartment if that's a very small
- space, right? Um-
- Yeah. I mean, there are other means, um, like a, a laser scanner, a
- Matterport, what you see in the picture-
- Yeah
- ... where you can, like, generate that small 3D model, uh, in a
- fraction of the time, um, way more flexible.
- Um, same for these little pipes, small
- holes. We're not an industrial drone which flies through that.
- As you said it, in the end we navigate visually
- and we take visual
- pictures. We don't have a laser scanner on board.
- So, um, if it's dark as it is on that middle picture,
- the drone will not fly. It will tell you.
- Yeah, yeah.
- On the other side, where Indoor Capture is great to use is in these
- big environments. So just, like, the rule of thumb here is
- anytime you need additional equipment to go to a different place, like a cherry
- picker, um, a scaffolding, climbers, uh, or where you l- need to
- get, like, laser scanners elevated, um,
- then Indoor Capture is great to use because the drone can
- simply fly in 3D
- space. It can fly
- up to your, in your warehouse rack, for example. It can fly under the roof.
- The camera turns 90 degrees upwards as well.
- There Indoor Capture is great, super fast, and
- cheaper than all the other means you see here, like a, like a cherry picker, for
- example. And you have it-
- Now that you mention... Yeah, sorry.
- Yeah.
- No.
- And you have it always with you. So, um, like, usually a cherry picker or stuff you
- have to order, you have to close the area.
- The drone is always there, it can simply fly there.
- And you mentioned something about the, uh, the angle.
- Um, so there was a question from Adam.
- "Can you go over gimbal angles and which to use based on the
- capture type? Does it know to capture above without
- having to run a different
- mission?"
- Yes. Um, that's a great question. I don't wanna get too technical
- here, but basically... So we did the exploration.
- The drone knows how the space, um, the structures look like,
- and then the drone
- works, it works in 3D... It has a 3D,
- uh, coordinate system. So if you imagine, like, the X, Y, and Z axis,
- it will start flying around the Z axis, so flying across the room shoo-
- shooting vertically at the walls.
- And then it will basically fly around the X and then the, the Y axis.
- So, and in the X and Y axis, since that covers the, um, the ceiling, for
- example, or, um, the floor or any other angle,
- the camera will automatically, um, turn to the angle,
- uh, which is facing the structure you're
- looking at.
- Yeah.
- And also it is a-
- And this is part of the... Sorry.
- Yeah.
- Go for it.
- And a- and also, um, our algorithm takes care
- that you don't capture
- everything three times like an X, Y, and Z.
- Mm-hmm.
- But it really,
- um, focuses, um...
- It determines the best angle of the surface, and then integrates it
- into the, the path where you can take the best
- image.
- Yeah. And this is the adaptive, a- adaptive flight planning that we do.
- Yeah.
- So, so Skydio does this for you, Adam.
- So it will do, you basically, like, define that boundaries, um, and the, the,
- the space. Sorry, define the, the asset that you wanna capture,
- and Skydio
- does the re-
- Yeah
- ... the, the, the rest for you. Um, we also have another,
- uh, for 2D capture we released last year a 2D
- lookup functionality. So you can also do 2D captures
- of, um, uh, you know, the underneath of a, of a, of a
- bridge deck, for example. Um, so we do have different,
- um, methodologies for different use cases, if that makes sense.
- Yeah, yeah.
- Um-
- The way how we set up Indoor Capture is you can...
- I mean, in our example here, we generated a digital replica of the
- warehouse. You can just say, "I use Indoor Capture only to capture the
- roof." And
- if you have, for example, big s-, um, steel truss, Indoor Capture will make
- sure always, like, to capture in the side of the steel truss
- to always have the
- right angle. Or you could use Indoor Capture just to
- cap- capture a large machine.
- Maybe it has to be dismantled or you want to, um,
- inspect the conditions on top
- of the machine. So you're flexible to use this tool how
- you want it.
- Yep.
- Um, I think we covered most of the, of the questions.
- Um,
- there, there is, uh, Eric is asking about the GSD.
- We did cover the GSD already.
- Um, it's one of the settings that you can change after the exploration,
- Eric.
- Um, so that will determine how close the drone
- flies, uh, to the object, so distance to surface.
- Um, and obviously will have a massive
- influence in the number of
- pictures and the time and batteries, um, necessary to capture
- that, um, that mission.
- Um, Andre, I think do you have any other
- comments, remarks, sentiments you wanna share before we,
- we wrap up?
- I mean, um,
- GSD, you see it here on the slide before your
- scan starts.
- Set up the distance-
- Yeah
- ... and you will see the, the resulting GSD, and it will-
- Yeah
- ... result in the time the scan will need.
- Um, I'm very open to questions from the
- audience.
- Yeah. I think we covered most of it, um, already, but we
- did-- we will share the webinar. So the, the link to the recording will be
- available on LinkedIn. But I've also shared, um, our webinar
- platform so you can register and have the, the recording there and access to
- many more webinars from Skydio. Um, any other questions
- I've missed, um, I'll be tagging Andre, and, uh, you can always
- reach out to me or anybody in the team.
- And, um, yeah, I think this was, this was great.
- Very informative. Um, Andre, any,
- any other-
- Thanks to everyone, uh, who listened.
- Yeah.
- Um, around the globe, we have a lot of colleagues from the
- States, um, I hope some
- colleagues from Europe. I know from some colleagues in Japan,
- so all around the
- world. Thanks for joining.
- As Alicia said, um, get back to us in case you want get--
- in case you want to... in case you want to get your hands on
- Indoor
- Capture or the other-
- Yes
- ... scan modes, and yeah. Thank you.
- Yeah. Thank you so much everyone for joining
- today.
- Um, Andre, it was such an amazing pleasure to have
- you here today.
- Actually, just a question came in from Anthony about
- repair-
- Yeah
- ... repeatability. So it's a great question.
- So Indoor Capture was basically just, uh, released, um,
- one month ago. Um,
- you can repeat scans currently. Like, you, you can repeat them.
- Um, you have to do the exploration again. You have to set it up again.
- Um,
- what we currently don't allow is that you can save a scan and the, the
- drone can simply repeat it.
- Um, but our overall vision for all scan modes,
- everything we do at
- Skydio is like to automate every process.
- So, um, like scan modes, like 2D GPS capture or map
- capture, you can save already. You can repeat it, and
- repeatability is very high up on our list.
- Um, so not supported right now, but
- maybe a talented team is working already on it.
- Hopefully coming soon. Hopefully coming soon.
- And with that, we will wrap up for today. But thank you, everyone.
- Um, thank you so much, Andre, again, and, uh, see you hopefully very, very soon.
- Bye, everyone.
- Thanks.
Watch Time
51 Mins 42 Secs