Transforming the way bridges are inspected with Skydio
Stantec uses the power of Skydio drones to overcome the challenges of traditional inspection methods that are slow, risky, and costly. The talk showcases the impressive results of an inspection conducted on the Mike O’Callaghan-Pat Tillman Bridge and introduces new inspection workflows that were able to cut inspection times and crews by up to 75%.
Michael Marshall, Bridge Engineer and UAS Program Lead at Stantec says that drone utilization significantly reduces inspection time and personnel requirements, while providing valuable data and operational benefits.
Learn more about Skydio for bridge inspection.
- [upbeat music]
- My name is Michael Marshall. I'm a professional engineer at Stantec.
- Been doing bridge and tunnel inspection for over 10 years
- now, and a little bit of
- asset management as well. I'm a SPRAT level one,
- so a rope access technician,
- and I'm also a drone pilot.
- Hi, Josh Sexton here. I sit in Bluefield, Virginia with Stantec.
- I'm a reality capture technology manager, and I also operate our
- global UAS program here at Stantec. So it's exciting to be here today.
- The project we are gonna be presenting here today is the inspection of the Mike
- O'Callaghan-Tillman Bridge, which is in front of the Hoover Dam.
- This was definitely the largest infrastructure project
- that I've done yet with
- Skydio. To set the context, our collective goal was to
- assess the value of autonomous drones of Skydio
- solutions for bridge inspection for the Stantec team, and the
- potential of integrating this technology from operational
- practices all the way up to more reliable and informed business decisions.
- [upbeat music]
- A traditional inspection is typically two people, a team leader
- and an assistant
- out on site, and they might have paper and maybe an iPad on site to take their
- notes. And so the end result is essentially a written report with a few
- photos on the end that's delivered to the owner of the bridge.
- But here in O'Callaghan-Tillman is an example of how we use some other
- techniques
- such as rope access, snooper trucks, maybe bucket trucks.
- And you can see here in this picture we have a rope access climber on the side
- there with a snooper truck helping us inspect the superstructure.
- So prior to probably 2017 or so, it was solely done with rope
- access and snooper work. The climbers would rappel down the columns
- and use
- what's called a belay technique and look at the arch, look at the columns,
- and then
- they would also rappel the columns individually.
- But around 2017, 2019, we started integrating in drones to
- the approach columns, and those were just the traditional manual drones
- looking
- up and down the approach columns to try and relieve some of that climbing that we
- had done for two weeks. That's kind of where it originated,
- and then we moved into
- the Skydio side.
- [upbeat music]
- QAS provides a wonderful platform when you don't need a hands-on inspection,
- and you're actually gathering a lot more data, and you can start comparing that
- data over time. So one thing that everyone wants to talk about
- is a digital
- transformation, and as part of our digital transformation,
- we're doing better data
- capture. It's capturing data. A lot of times maybe we don't need
- climbers or bucket trucks for hands-on access.
- Maybe it's just a routine survey, so we can take a first pass with a
- UAS. And something that's often not referenced
- is drones are a
- great equalizer too. You know, maybe we have staff with physical
- limitations, a
- drone levels the playing field for a lot of those individuals.
- It takes our load off our staff, it makes their job generally easier,
- and then
- it also is able to involve more staff.
- So it helps us leverage our staff better, and then it also promotes those people
- that may have physical limitations.
- It promotes their career because this is another tool that they have so they can
- be
- engaged. So it's just a win-win for us.
- [upbeat music]
- You know, we're real happy with Skydio.
- We've been flying drones at Stantec since 2013.
- We purchased our first drone. We actually had a full global
- governance by 2016, so we have a lot of stick time with our
- 70 pilots globally with drones, but those have all been manual drones.
- And we're seeing a big push in the inspection market, specifically
- with transportation and the built environment,
- that we're leveraging this
- technology in new ways. And it makes sense for us to have our
- engineers behind
- the sticks because they know what they're looking for, and this is a tool
- for them to have aerial inspection, limited access, hard to reach areas,
- and they can focus their efforts in getting there.
- And maybe they don't have to be in a bucket truck or with rope access.
- It lowers congestion. And with the autonomous drones,
- the drones are
- getting smaller, the footprint's smaller, they're easier, they're more portable.
- You can carry them on an airplane now, and the learning curve is much
- lower with a Skydio drone we found versus our typical inspection rig,
- which is a Falcon eight, which is also a very good drone,
- but it takes a much
- higher level of training, and then it still doesn't have the
- obstacle avoidance
- that the Skydio does. So for a lot of routine inspection,
- the Skydio just
- makes sense just because of that built-in object avoidance, 3D
- scan, all the autonomy that it offers.
- It's just a fantastic platform for this type of inspection.
- The one thing I'll add there is with the traditional drones,
- typically you have a
- pilot and also someone who's the visual observer, possibly operating some of the
- camera controls. So there's a lot of folks on site
- that you're using, plus the
- engineer that's also looking over how is the data capture being done.
- With Skydio, we've been able to consolidate a lot of
- that, especially with 3D scan,
- into one single person, the engineer, the 3D scan, and the data collection all
- into one hat essentially, and huge game changer.
- [upbeat music]
- Traditionally, before we brought drones in,
- this O'Callaghan-Tillman project would
- take around two weeks of at least eight climbers,
- and then you have all the rest of
- the staff on board as well. Operators for the snooper trucks,
- three to four of
- those, traffic control managers, at least four, possibly more,
- and an officer.
- And of course, there's more of the team on site.
- But I would say at a minimum, that's what you're looking at for two
- weeks.
- By just integrating drones, we're able to bring
- that down to just one week.
- We keep that same staff of eight and then the operators and such,
- but we're able to
- bring in a drone team and just look at part of the bridge
- and completely reduce
- the staff down to, say, a drone team of three or so.
- And one week of drone work eliminates a week of climbing work.
- But then when we were bringing in Skydio, the obstacle avoidance and other things
- that we had just talked about, it also reduced that down to just a couple
- of days.
- Traditionally, we've been able to use manual drones to
- look at corners of the
- columns and go up and down and use video, but with Skydio, we are able to do that
- exact same data capture in just two and a half days. And why was that?
- The pilot did not have the fear of crashing into the bridge.
- Even though we would lose GPS reception when you get up to the top
- and kind of tuck
- yourself into that steel girder, you're still in full control
- and object avoidance,
- so you're not in that fear of hitting the bridge,
- and you can get close to capture
- that data. So with that, the data capture was much quicker.
- In the terms of this project, when you take away
- that amount of labor force and
- hours throughout the week, you can take it down from,
- let's say a $200,000 cost in
- two weeks down to about 105,000 in a week.
- So that alone helps you justify the purchase of a drone
- or some of the software
- that comes on board with it. [upbeat music]
- With the traffic control, it's a major operation.
- There's a lot of folks involved in a survey like this, especially a big
- one like Ocala and Tillman. There's a lot of congestion,
- and everything we do, we
- try to do safely. We're taking the load off of our
- individuals.
- They're not having to climb. We're able to access those assets from a
- safe spot
- on the ground, and we're not creating traffic congestion
- because we don't have to
- have those folks on the deck blocking a lane of traffic.
- Just the impact to the operations of the highway in this instance are just
- huge.
- The one thing that we're looking for is making sure we have the bridge engineer
- on
- site during the data capture, right?
- So they're reviewing not just the pictures and the video being taken, but also the
- 3D scan. When something might be seen, we can at least relay that back to the
- rope access technicians and say, "Hey, we need to actually rappel this and take a
- look at a certain location." And that actually saves time as well.
- We can hone in the exact locations where we need hands-on access to and
- not necessarily just drone everything, 'cause in the end, hands-on access
- is still
- part of our industry, but the drone helps us key in on those
- certain locations that
- we might wanna have more attention to.
- The drone can actually get to the spots that rope access or bucket trucks have a
- hard time getting to. So if they see something
- that they can't easily
- assess, you can send the drone in to take a closer
- look for them.
- I had experience with the other drone work that we've done on site.
- Having a week-
- Mm
- ... long time, it... You have to account for the wind
- conditions and things that
- we're gonna go over, and I expected things to take a lot
- longer.
- But when we started using 3D scan, things we're gonna see here,
- you can immediately
- see we're collecting this data rapidly and accurately as well.
- We actually collected more data in a shorter amount of time with Skydio.
- It was a great-
- Yeah
- ...
- experience. [upbeat music] The first thing I look up
- whenever we get a prospective
- bridge to inspect with a drone, it's airspace, right?
- So in this case of Ocala and Tillman, it was part of the airspace was federally
- restricted, so we had to get that permissions done.
- But also when you're looking at maybe Class B airspace and getting closer to
- airports, if you're having to fly in those
- environments, you need to start looking
- at waivers and permits to, to actually fly in those environments.
- But the other big thing with bridges is that you can usually not account for
- is the
- weather. Wind, you can look at weather reports all day,
- but bridges create this
- tumbling effect, and also the environment around them
- create gusting and things
- that could go wrong with drones. So having that handheld on site
- and understanding
- your weather conditions is really, really important.
- The bridge is there for a reason. Unless you have an underpass underneath
- the
- bridge, typically the terrain under a bridge
- is not suitable for a
- helipad, for example, for a takeoff and landing zone.
- So you need to account for where are we gonna take off
- and safely land this drone,
- uh, and also be able to keep our visual line
- of sight to check that box off and be
- able to do all of our work efficiently.
- [upbeat music]
- Prior to Skydio, I had some experience with other types of
- drones on bridges, and I
- will say the first thing I noticed putting a conventional drone up underneath a
- bridge, the GPS loss and, of course, that tunneling effect we just talked
- about, it
- made it really uneasy. And you are focusing more on the drone than you
- are the
- bridge inspection in some cases 'cause you're focused on not hitting the bridge
- and
- losing all of your data. But with Skydio, if we lose signal in any case,
- it is able to come back to us and also use objective winds to come back
- to us.
- And on the flip side of that, with the manual drones on a bridge site,
- if it's, you
- know, has a pier or something in the way
- when it's on its way back, it may not
- avoid that path. Or let's say that path is impeded, even if it follows the same
- path how it got there, it may have a crash.
- But with the Skydio, it's at least able to do object avoidance,
- and you can help
- customize where you want that safe zone to be.
- Yeah, the lost link procedure is better with the Skydio solution,
- and that's one
- reason why we like it, is because with enterprise autonomy,
- it can
- navigate based off vision instead of GPS for returning to home, and that's a
- huge safety margin we enjoy, you know, and that's one reason why we
- deploy the solution.
- [upbeat music]
- We have to understand that bridge inspection traditionally
- is a written report with
- photos, so that's at least what we need to get our
- data into at some point in
- time. But we're migrating more towards the
- digital world, right?
- So we're moving more towards a digital twin, and how are we gonna deliver that to
- the client? Some clients might not have access to the
- 3D modeling world or
- have the ability to process some of this data.
- So how will we get that data to them and essentially deliver the report in that
- digital package? With the video, there, there's a time and a place for it, right?
- If you just wanna look at a specific detail, we can go and take a look at that,
- take a video, take a still photo, and integrate that in the report quite
- easily.
- But the drawback to that is unless you have a scale right on the
- structure
- or very close by, we're not able to put a scale to what
- you're looking at in the
- video. You're also not able to measure or annotate ba- right on top of that.
- The other drawback is you're at the mercy of how
- that video was captured or how the
- photo was captured, so you're not able to see maybe the
- environment or what's going
- on around it. One example of this is manual capture.
- You might see some staining on your pier or something like that, and you might
- write it down as, "Oh, we found staining through this video,"
- but that might be
- caused by a drain up above draining right directly onto the pier below.
- So that's where a model could come into play.
- We can see the causes of the defects or the things that we're seeing and not
- just what is the defect in the video.
- What the model does give you is that actionable visual intelligence that
- Michael was talking about. It gives you perspective.
- But going into that, if you're gonna create a model,
- you have to realize, are you
- doing it once or are you doing it multiple times?
- If you want the models to overlay, you really have to think hard about ground
- control. If you're going to measure off that,
- you have to think about your GSD
- and your ground control plan to accurately align that model and be able to defend
- it. Those are all things that we need to think about going into a
- survey.
- Are we just looking at the structure and looking for defects, or do we wanna
- monitor the structure over time and use machine learning to map
- cracks and then see how they're propagating from year to year?
- Have to have a good idea of your deliverable going in.
- It's not just go out and capture data and figure it out on the back end.
- [upbeat music]
- 3D scan is the software that sits on our enterprise, uh,
- subscription, and it can visually assess or
- explore a specified structure. So you create a binding
- box, and then autonomously generates and executes a
- flight path to ensure high data quality is
- captured, and then that can be used for inspection and 3D
- mapping or 3D modeling. So it basically takes away the
- need for any operators to have any previous experience on
- photogrammetry. You are literally telling the drone, "This
- is what I want to
- capture," and then the rest, Skydio 3D scan does it for you.
- The only thing I'd add to that for bridge inspection
- is before you're maneuvering
- the sticks around or maneuvering the drone around, here we're able to tell it,
- "Here's my object that we're looking for.
- Go get me the data that I need to do a 3D model." And
- while we're doing that, I can
- also view the screen without having to fly and see exactly what's going on with the
- bridge and be able to look for those defects and things like that
- and really focus
- on the job at hand. That's, is bridge inspection
- You also have to consider what asset you're looking at.
- Concrete is very forgiving for-
- Yep
- ... photogrammetry, but steel girders, for instance, or if you're monitoring
- an area that has a homogenous capture area, like a
- forest or sand and so forth, you have to really increase that overlap
- front and side. So you do have to have a good basic
- understanding
- of photogrammetry to get a deliverable that's suitable for your
- process. Versus a manual drone, you're not trying to estimate, do I
- have 70/70, 80/80? Which you generally just take too much,
- and that's
- the beauty of the Skydio, the fact that you do set the front
- and side overlap, and
- then it takes care of the rest. And it keeps the GSD consistent, which I
- think shows that helps with the stitching process immensely.
- You're able to see, one, how is it being captured
- and what's the total picture of
- it. But what I really liked is, you know, I'm not able to necessarily be on every
- single controller, so when it comes back, we're able to see that model at the very
- end and at least view the data as it came in, and see that before we say, "Okay, we
- have the information we need." Or the bridge inspector can at least look
- at it and
- say, "We're not seeing any hot button issues
- that we need to look at with the rope
- access." So yeah, that was really helpful.
- So when we took a look at maybe a two-span structure, such as the one in the video
- here, typically I've been able to knock those
- out in about 10 minutes of 2D
- scan up, being able to take it back and it stitches in less than an hour.
- Now, not only do we have the information back
- from what does the superstructure
- look like, what does the bottom of the deck look
- like, but we also have the ability
- to take that model and then compare it over time quite
- easily.
- In terms of 3D scan, it's rather quick, but even faster is just to do that
- 2D up and be able to come back from the field with images that
- are right up against
- the superstructure. So again, looking at that ground-level inspection
- that we're
- talking about, if I'm just walking underneath the
- structure and looking up, we can
- see quite a bit. But when we take a drone
- and put it maybe three feet below the
- superstructure and we're taking pictures and modeling that, now we're able to see
- even more and take that data back and be able to compare it over time.
- This is the GenX platform. So this is one of the cloud solutions
- that we processed
- this project in. We're looking at the O'Callahan-Tillman
- Bridge here.
- The nice thing about this, it puts everything in perspective.
- We do have some areas highlighted.
- This is a very new bridge, so it's really hard to find distinct
- features and
- failures on this bridge. There weren't any.
- But we just drew some boxes around here so you could see representatively,
- if there were problems, how you could annotate those and so forth.
- But it gives you a lot of perspective on the ability to see
- the whole asset at one time. And the fact that photogrammetry is
- getting so good that it can process it as one complete job, not
- individual peers, that was just a major breakthrough for us,
- and we were really
- happy. And a note on this, we did fly this at a fairly coarse
- resolution.
- Alicia, I think it was, what, 24 feet approximately?
- The GSD was a little coarser, but we were showing a pilot test.
- You know, we weren't absolute accuracy on this,
- mining it with AI for crack
- detection. This is just capturing the asset a-
- and showing the possibility of the
- technology.
- Do you recall the number of pictures, like the total number of pictures, Josh,
- for
- the entire data sets?
- I think it was around 9,000.
- 9,000 pictures. Yeah. It was processing less than, I think it
- was about 20 hours,
- something like that.
- Yeah.
- So again, in a day and a half, you could be potentially training the
- team that needs to go next in the fields.
- You could be communicating much better across different departments.
- Uh, you know, there's so much value in, in having that digital twin of your
- infrastructure.
- The one thing that I really liked about the Skydio drone with the 3D scan is it's
- able to tuck itself in between the column and the rock walls that you're seeing.
- Super beneficial. It's able to do that without the fear of hitting,
- let's say,
- you know, the pier or a piece of the rock
- and then falling to the ground.
- And you can see in Josh's example, we actually did find a drone at the base
- of one
- of the piers, possibly a casualty of that exact thing.
- One thing that was amazing was just watching
- that Skydio go around and get the full
- context of what that column is, not just staying away from the rock wall,
- but the
- entire capture, and also turning around
- and looking at that rock wall so we
- can find defects that are possibly associated with that
- that could cause us
- problems with the bridge. So quite amazing.
- The other thing Josh is kinda showing here, too, is
- when we can click on the model
- there in GenX, we're able to also look at every photo
- that was taken of a specific
- location, and this was really beneficial for the inspection.
- So now we're back in the office, we're looking through what did we find
- here on
- this bridge, and reviewing our data.
- We're able to quickly move around, as Josh has shown,
- and look at very specific
- locations that maybe we have interest on, and annotate right on top of
- the model or on top of the picture itself.
- The other thing I wanna note here is, you know, we have some video that actually
- came out of this as well as a test run. We're able to put in that survey as well.
- So we can stack different surveys and different items on top of this
- and get a total inspection right out of this platform.
- With traditional bridges, yes, you would have to stop traffic.
- If you were to do a total 3D scan and just let the drone do everything,
- you would
- definitely have to stop traffic, 'cause it would be up above the deck
- and possibly
- in the live traffic lane. But one thing that we did here on
- O'Callahan-Tillman was
- we kinda divided things up and then stitched it all together at the
- end.
- So every column was essentially its own subject of a 3D scan, and you
- can use this on other bridges as well.
- You can look at just a substructure and say, "Let's scan the substructure.
- Maybe let's use 2D scan up to do the superstructure,
- and then maybe we do a manual
- flight off to the side of the deck if we don't have those traffic closures to
- stay
- away from live traffic." So instead of thinking of it as, "Let's scan the whole
- bridge," maybe, "Let's scan the components
- and then bring all of our data in
- together at the end."
- It's part of our procedure. We don't fly over people or live traffic.
- You know, we would take those with obliques off to
- the side.
- And a lot of DOTs are actually inspecting the deck of those
- bridges with
- terrestrial vehicles and systems like that.
- The primary focus, I think, of a lot of Michael's work,
- and he can speak to this
- specifically, is the underside of the superstructure,
- the piers, and that sort of
- thing, not so much the deck.
- One thing that we're starting to vet in QA is essentially the crack detection
- softwares that are coming out. So yes, those are starting to come to the
- forefront
- of our inspection softwares, but I would say in this case, we
- were more concerned
- with trying to vet what the notes were from previously and be able to carry
- those forward. So yes, we're able to see those cracks
- and possibly measure the
- lengths. The widths, unless you have a detailed scale or, uh,
- possibly using AI
- detection, not able to do that exact width very easily.
- Yeah. And that's where you have to be really
- careful, too.
- That's where you really wanna make sure your GSD is correct
- and your ground control
- is in place so you can actually defend that data.
- And, you know, that would be the workflow for that.
- Another thing that is great about the 3D model
- is the textured mesh.
- So if you have an area that is eroding, you can actually
- see that depth in three dimensions versus a 2D photo,
- where you may not get the idea of that depth.
- It does have a volume behind it, which is nice.
- This project, I tried to look back on and find ways that we can improve,
- and it was
- quite difficult to do from the drone side. It was very seamless.
- A lot of the data was captured in maybe even a day
- and a half, and the extra day in
- there was to make sure we buttoned things up and had what we needed before we left.
- I would say the sheer shock of seeing how quick drones can be integrated with a
- bridge inspection program, the type of data that comes back
- is really beneficial.
- You know, there's lots of hurdles to think about,
- where they can be deployed and
- how they can be deployed, but when they are,
- it's an absolute game changer in
- bridge inspection.
- I think UAS data capture is going to be extremely important as
- all business units move into digital transformation.
- Being able to safely compare over time, I think we're gonna see a move away from
- static reports and move into that digital intelligence
- mindset, creating a common operational picture
- where we can be better storytellers
- to our clients, and being able to have near real-time
- asset monitoring
- following events, moving into Skydio Cloud.
- You know, though the one caveat I will say,
- and I think it's been brought up
- several times, is you do have to have a robust system to
- support your pilots, make
- sure you're doing everything safely, that you're following the law, and
- that you're
- meeting all the requirements for your client and airspace that you're flying in.
- It's very important to do things the right way, and that's one thing we're really
- proud of, and that's one thing that Skydio also offers with their
- solutions,
- which is great.
Watch Time
20 Mins 32 Secs