Latest / CLEs You Actually Want to Hear / Early Resolution: How Strategic Use of Visualization and Animation Can Help Your Client Avoid Trial with Charles Fox
Transcript
- 0:03Good afternoon, and welcome to Early Resolution.
- 0:07How strategic use of visualization and animation
- 0:10can help your client avoid trial. I'm Lori Kripman,
- 0:15the vice chair of the Beverly Hills Bar Association's
- 0:18litigation section. I'm a certified family law
- 0:20specialist. I'm very happy to introduce today's
- 0:23speaker, Dr. Charles Fox, who goes by Chuck.
- 0:27Chuck is the Senior Director of Client Development
- 0:30and Engineering Services, Incorporated. After
- 0:34earning a doctorate at Ohio State University
- 0:38and completing a neuroscience fellowship at the
- 0:41University of Michigan, Chuck made a bold pivot
- 0:44in 1994, leaving academia to join a pioneering
- 0:49team focused on 3D animation and scientific visualization.
- 0:55What Chuck started as a temporary project blossomed
- 0:59into three decades of pioneering work at the
- 1:03crossroads of science, technology, and storytelling.
- 1:08Over the years, Chuck has guided the multidisciplinary
- 1:11teams in creating visualization tools that bring
- 1:15clarity to complex scientific ideas and shed
- 1:18light on unseen events. Assets that prove invaluable
- 1:24in mediations, arbitrations, and trials. We can't
- 1:27wait to see what he has to share with us today.
- 1:30Chuck, the floor is yours. Oh my gosh. Thank
- 1:33you so much, Lori. I realized in your introduction
- 1:37that 1994 sounds like a long time ago. And it
- 1:41doesn't feel like it was that long ago, but...
- 1:45Things have changed dramatically. And I'm going
- 1:47to talk about some of that today, especially
- 1:50in association with this technology that that
- 1:55I'm going to share. So let's pop up my PowerPoint
- 1:58here. And I've managed to do it twice. Are you
- 2:01seeing my screen right now? Yes. Awesome. So
- 2:07I have a slide here called storytelling versus
- 2:10reality finding and. Back in the early days,
- 2:17we talked about 1994, and if you go back to the
- 2:211990s and early 2000s for that matter, when we
- 2:26began creating 3D animation for the courtroom,
- 2:31we really viewed it as a storytelling tool. When
- 2:34engineers were testifying about cases in the
- 2:37courtroom, there were often complex topics. If
- 2:41it was a vehicle accident case, there were issues
- 2:43like braking distance and line of sight and things
- 2:47like that that were really difficult to describe
- 2:50and sometimes not captured by fact -finders.
- 2:55And using 3D animation as a tool for putting
- 2:58those concepts on screen for putting fact -finders
- 3:03in the driver's seat of a car or a witness view
- 3:07for an industrial accident or perhaps a crime
- 3:11scene was really valuable because these concepts
- 3:14really transcended the technical jargon that
- 3:17the expert was likely to use. So it allowed us
- 3:20to really teach these complicated concepts to
- 3:24really powerfully testify on their own. And we
- 3:28also Learn that these are quite memorable that
- 3:31in fact when you put this kind of imagery in
- 3:34front in front of someone they can often recall
- 3:36details from it much longer than they can recall
- 3:39details if they were to read about it or hear
- 3:42about it as part of testimony so. The tools had
- 3:46tremendous value at this stage as a storytelling
- 3:50tool, but they weren't necessarily Reality -finding
- 3:53tools in some sense. I always you always bristled
- 3:56at the term cartoon, but in some sense at that
- 3:59stage they they were Now I think if you think
- 4:04about sort of 2015 is kind of a landmark year
- 4:08in my mind. It's it's somewhere in that region
- 4:10things began to shift toward reality finding
- 4:13and and What do I mean by that? Well, what I
- 4:18mean by reality finding is really the tools that
- 4:21were used to collect data that we were building
- 4:25the animation in became very reliable. So before
- 4:29this time, most of the time at an accident scene,
- 4:32we relied on measurements that were made like
- 4:35you might imagine with a tape measure or a measuring
- 4:40wheel on a roadway and perhaps a you know, policeman's
- 4:44accident report or a drawing of an intersection
- 4:47that may or may not be accurate. In 2015, we
- 4:51started seeing laser scanners in the space, and
- 4:55we started seeing laser scan data on a fairly
- 4:58regular basis. And so instead of maybe a handful
- 5:02of measurements from a scene, we had millions
- 5:04of measurements from a scene. and they were accurate
- 5:08down to the millimeter. So we could understand
- 5:13the geometry of a scene in a high level of detail
- 5:16and also it was highly reliable. I'm going to
- 5:21talk in a little bit about kind of defining space
- 5:24as the first step. And this was the first step
- 5:27toward something I would call reality finding,
- 5:30that we could reliably take data and say, hey,
- 5:33we know this vehicle moved this distance in this
- 5:37period of time, and we know precisely what those
- 5:39distances were and what the details were about
- 5:42that particular scene. As you glean from the
- 5:46title of this presentation, this has a lot to
- 5:50do with the ability of these tools to help clients
- 5:53avoid trial, because rather than explaining something
- 5:57in the courtroom to a fact -finder and having
- 5:59a cartoon to do it better, these started to become
- 6:03really reliable depictions of what happened in
- 6:06a particular event. And when we do that well,
- 6:10and the story is clear, It often can drive settlement
- 6:13early on in a case. So capturing space. What
- 6:21do I mean when I talk about capturing space?
- 6:26Well, this is an example of the early days of
- 6:29capturing space. And this actually predates the
- 6:32laser scanners in 2015, but it was something
- 6:35we could do on a large scale. And what you're
- 6:38seeing on this screen is on the left side, you're
- 6:41seeing a grid, essentially like a piece of graph
- 6:45paper. but it's a complicated shape. It looks
- 6:48like it's draped over something. And in fact,
- 6:51what that is, is United States Geological Survey
- 6:54data for Mount St. Helens. So this is three meter
- 7:00data. So if you imagine a grid where the points
- 7:03are spaced by three meters, we have XY coordinates.
- 7:08So we know basically on a north -south, east
- 7:11-west, grid, where that point is located. And
- 7:15then we have elevation data, z data, which tells
- 7:18us how high that point is. And when we know three
- 7:21points, we actually know where that point is
- 7:23in space. And if we put a bunch of them together,
- 7:26we can create a grid like this that is the shape
- 7:30of, in this case, Mount St. Helens. On the right,
- 7:34we have aerial photography from Mount St. Helens.
- 7:38And if we combine those two things, if we map
- 7:41that color data to the grid, we get a really
- 7:46nice 3D depiction of Mount St. Helens. And this
- 7:50is useful, for example. In an aviation case,
- 7:53if we wanted to have an airplane flying by Mount
- 7:56St. Helens and have a view from the airplane,
- 7:59we've essentially got a digital twin of Mount
- 8:01St. Helens that we can look at from any vantage
- 8:04point that we desire. So that was some of the
- 8:08first sort of three -dimensional data we began
- 8:11to work with that had some kind of deep scientific
- 8:17basis in terms of being able to pull out data
- 8:20from North America. Laser scanning works on the
- 8:25same principle. We're capturing points that have
- 8:28x, y, and z values to them, but we're doing it
- 8:32at much higher detail. So I mentioned that the
- 8:36USGS data is captured at three meter intervals.
- 8:40The scanners that are on the screen right now
- 8:43capture data on inches to millimeter intervals,
- 8:48or in the case of the scanner on the right side
- 8:51of the screen that's in Nick's hand here, down
- 8:54to 40 microns spacing. So extraordinarily detailed
- 8:59data. On the left is a ferro scanner, and if
- 9:02you're a litigator and you've worked on accident
- 9:05reconstruction cases, you've seen data from a
- 9:08ferro laser scanner. These are very common. Law
- 9:10enforcement uses them a lot. We have a fleet
- 9:13of them that we take into the field, and you
- 9:15can imagine surveying a scene. Well, a surveyor
- 9:20is taking points one at a time. The Ferro Focus
- 9:24takes points millions at a time. In about a six
- 9:27-minute scan, we get millions of points in a
- 9:31volume that's about 300 meters or so in diameter.
- 9:39So you can imagine having this large volume of
- 9:42points. that are millimeter interval points,
- 9:46it's a pretty detailed description of a geometric
- 9:49space, perhaps where an accident occurred. The
- 9:52scanner that David's wearing, the guy in the
- 9:54red shirt, is a similar technology. It's a laser
- 9:58scanner. It's actually... two laser scanners,
- 10:01there's one above his head and one in front of
- 10:03where his hands are, that are oriented orthogonally.
- 10:07They're one scanning horizontal and one scanning
- 10:09vertically. And he can walk that scanner through
- 10:12a space. And so what's really handy about that
- 10:14is in a complicated space, he can capture like
- 10:19the interior of a building or industrial space
- 10:21fairly quickly by being able to move through
- 10:23it. The hand scanner I mentioned Nick is using
- 10:27captures very, very high detail data. So if we
- 10:30need a small area captured at very high detail,
- 10:34we would use a hand scanner and we use those
- 10:36in the laboratory or in the field. And then finally,
- 10:39you can tell I added this one later, there are
- 10:43vehicle mounted scanners now that allow us to
- 10:46scan roadways in fairly high detail. And what
- 10:49makes that really important is roadways are a
- 10:53dangerous place to scan and often we can't shut
- 10:56down traffic. We might be at a scene months after
- 11:00an accident happened and one of the ways we can
- 11:03scan the roadway safely is by driving through
- 11:06it and we can make multiple passes with a vehicle
- 11:09and capture a roadway in a fairly high level
- 11:12of detail. So this is what that data looks like.
- 11:17This is the picture on the right hand side of
- 11:19this screen is a scan that David did in a industrial
- 11:23space. And you can see really what I'm talking
- 11:27about in terms of the level of detail. It captures.
- 11:30not only XYZ information, but it also captures
- 11:34color information. So a term that I'll use is
- 11:38point cloud. And the point cloud just refers
- 11:41to the collection of these 3D points, each of
- 11:44them having XYZ information. And the point cloud
- 11:47is colored, so each of the points also has color
- 11:50information. So it gives a pretty good depiction
- 11:54of that three -dimensional space. One of the
- 11:58things I'll mention about this slide that you
- 12:00should be aware of, too, in terms of when we
- 12:02collect data, we can use multiple forms of the
- 12:06scanner at the same time. So if you look in the
- 12:08background of this scan, if you look really carefully
- 12:11in the back right side of the scan, you see a
- 12:14ferro scanner in the background on a tripod.
- 12:16And that's because we scanned this space with
- 12:18both scanner types because we wanted a really
- 12:23high fidelity scan for this. Both scanners give
- 12:27us a different view on that data that increases
- 12:31its fidelity. This slide is showing data that
- 12:37was taken with a laser scanner that was on a
- 12:39drone. So we can actually fly laser scanners
- 12:42over roadways as well. And this is a roadway
- 12:47that was captured with a laser scanner. There
- 12:51can be challenges. We can't overfly roadways
- 12:54directly. We have to fly off to the sides for
- 12:57safety reasons. And if it was a really busy road,
- 13:00we would be very cautious about putting a drone
- 13:02up in those areas too. There can also be areas
- 13:06that we can't access with drones for a variety
- 13:10of reasons. So we would drive those in those
- 13:13cases. And so one of the things that you'll notice
- 13:18in this one, a little pro tip in the background,
- 13:21is you see in the lower part of this image of
- 13:24the roadway a checkerboard marker. It's a white
- 13:26and black checkerboard. And those are used because
- 13:32we can map very specifically with GPS tools,
- 13:37where those... markers are positioned, and then
- 13:41we can map this data back into, for example,
- 13:45USGS data or other terrain data or GS data that
- 13:50might be relevant to a particular case. So it
- 13:53allows us to combine data sets. And as you're
- 13:56going to see as this presentation proceeds, a
- 14:00lot of the secret sauce of really doing this
- 14:03well has to do with combining data sets. This
- 14:09is just a shot of a scanner in action. This is
- 14:14a Ferro scanner. We're at a This was a refinery
- 14:20fire explosion and fire that happened. It involved
- 14:24a really large area of damage and destruction
- 14:28that we needed to capture. So you think about
- 14:323D models and where they come from in scenes
- 14:35that are chaotic like an explosion or a fire.
- 14:39Laser scanners are invaluable because we can
- 14:41capture that 3D chaos in a really detailed fashion
- 14:46and some. you know, engineering drawing or diagram
- 14:50of the refinery, while it might be great in telling
- 14:54us what the state of the refinery was before
- 14:56the explosion, if we want to know what it is
- 14:59after the explosion, we need to scan it. And
- 15:02in fact, we'll frequently do these scans sequentially,
- 15:06so as cleanup is occurring at an event like this
- 15:10and evidence is being... unearthed in essence
- 15:14or revealed as a cleanup happens, we can actually
- 15:17capture that with scans that are going on day
- 15:20after day. This is what that data looks like
- 15:23when you scan an entire area of a large scale
- 15:27thing like this refinery. This is just a small
- 15:31part of the refinery, but it involves hundreds
- 15:33of scans that are stitched together to get this
- 15:37particular point cloud. It's not like we pop
- 15:40the scanner out there in a large scene once push
- 15:43a button and walk away. There is some skill to
- 15:46choosing where you place the scanner so that
- 15:49your scans are overlapping and you're doing the
- 15:52right thing to capture this. But the super cool
- 15:55thing about this technology is once you have
- 15:58this XYZ plus color data, you can enter it. And
- 16:03you can go into the data and look at it from
- 16:06any vantage point that you wish, being, you know,
- 16:10any scale that you want. So human scale is the
- 16:14most frequent and the most important. And so
- 16:17here we are standing at the site of this explosion
- 16:21on a particular day of the cleanup. And we have
- 16:25everything precisely mapped in here. It's not
- 16:29just a picture. This is a 3D point cloud. So
- 16:31we can measure anything in here you want to measure,
- 16:35down to millimeter accuracy. So if one of the
- 16:38experts decides, hey, I want to measure or how
- 16:41far this piece of debris was from this point
- 16:43of origin, we can do that for them. And if they
- 16:46want to do it six months or a year after they
- 16:50were boots on the ground, we can do that. We've
- 16:53got it on a server, and we can pop it up and
- 16:56go back to that site again and again. When you
- 17:00saw that image, you sort of got a sense it's
- 17:02like VR, and we often go into it in VR. So this
- 17:06is just a movie. in one of the labs in that data
- 17:10set. So we can put on a VR headset, we can get
- 17:14in that scene and we can walk around it, inspecting
- 17:19it digitally, if you will. It's not interactive
- 17:23at this point, we can't go in and move things
- 17:25around, but we can certainly dive into that data
- 17:29and get a one -to -one view of it. So I'm gonna
- 17:35take a breath at this point. I know Suzanne was
- 17:37screening questions. Are there any questions
- 17:41at this point? Are the slides looking okay? Technically,
- 17:44are we doing well? Doing great. No questions
- 17:46so far. Excellent. So what I'm going to do now
- 17:50is shift gears from kind of some of the background
- 17:54to a case study. So you can kind of understand
- 17:58how this all comes together and why it's so vitally
- 18:02important to capture that data up front and why
- 18:05I spent all that time telling you about it. So
- 18:08in this particular accident, it involved a powered
- 18:12concrete buggy and so these these machines you
- 18:15may have seen them they're sort of like a wheelbarrow
- 18:18if you put a motor and tracks on it and you can
- 18:21move like big things that you couldn't normally
- 18:24move in a wheelbarrow like big chunks of concrete
- 18:27or big rocks and things like that and this was
- 18:29a this was a site where landscaping was occurring
- 18:32and they were tearing out a section of sidewalk
- 18:35at a location. And this worker was moving chunks
- 18:39of the concrete away from the construction site
- 18:42with this concrete buggy. They were loading the
- 18:45concrete then into a skid loader that then lifted
- 18:49it up and put it in the back of a dump truck.
- 18:51And so he was operating this powered buggy between
- 18:54the skid loader and the dump truck. And you can
- 18:56kind of see that in this image here. So... Typically,
- 19:02when my team gets involved, it's sometime after
- 19:05the accident occurred. Rarely are we there at
- 19:08the time of the accident. We're often there days,
- 19:13months, sometimes even years after an accident
- 19:16has happened. So capturing the scene while the
- 19:24value might diminish a little bit. You know,
- 19:28what I'm gonna show you is it's still absolutely
- 19:30vitally important and really valuable to capture
- 19:35the scene. So this illustration shows the collection
- 19:40of things that were put together to reconstruct
- 19:43the scene at the time of the accident. This is
- 19:46the, if you look up at the top left, you'll see
- 19:49something labeled B, sight scan. And you recognize
- 19:53by now that's a point cloud. That is point cloud
- 19:56data from the day that we laser scan the scene.
- 19:59And you see the dump truck isn't there, the skid
- 20:01loader isn't there, concrete buggy isn't there.
- 20:03In fact, there's a minivan parked roughly where
- 20:06the truck was. That's okay. We've got a phenomenal
- 20:12set of 3D data here that we can work with to
- 20:15rebuild this scene. We, in fact, had to go to
- 20:18another site where the concrete buggy was stored
- 20:21to actually laser scan the concrete buggy, which
- 20:24we did. That's right here as C. D is the dump
- 20:28truck, and this is the actual dump truck that
- 20:30was involved in the accident. We were able to
- 20:31go out and scan that. It had been in use since
- 20:34the time of the accident, but it was unaltered
- 20:37from the accident itself. And then we actually
- 20:40had some manufacturers' data just regular old
- 20:45engineering CAD for a skid loader. So we didn't
- 20:47have to really laser scan that. We could just
- 20:50put the skid loader that we had into this 3D
- 20:53scene. So we need a 3D scene that will actually
- 20:58depict the scene at the time of the accident.
- 21:01And that's the exercise we go through with all
- 21:04these objects, with the sight scan being the
- 21:07key thing that's gonna tie all these elements
- 21:10together. And you're thinking, well Chuck, how
- 21:13do these things all tied together if they're
- 21:16months later? The key is first responder photos
- 21:21or witness photos or surveillance video or this
- 21:26thing that's really ubiquitous these days everywhere
- 21:29we go, we're on camera. And in this particular
- 21:32case, first responders took two photos that were
- 21:36absolutely critical for us. And what you're seeing
- 21:39in this slide is on the left side of this slide
- 21:43is the skid loader. On the left side of the left
- 21:48side A is the accident photo. This is the photo
- 21:51that the first responder took at the time of
- 21:54the accident. The right side on the right side
- 21:57of the slash on the left panel is the computer
- 22:01model of the skid loader that's now placed exactly
- 22:06where the skid loader was in the photograph.
- 22:09Similarly, there was a picture of the back of
- 22:12the dump truck with the victim of the accident
- 22:14on the ground, and the left side of that is the
- 22:17photo, the right side is the reconstruction of
- 22:20the back of the truck. The technique that's used
- 22:24is called photogrammetry, and what we can do
- 22:29is simulate the camera that took the photos in
- 22:32the computer move the computerized camera to
- 22:36the exact location where the camera was that
- 22:39took the photo, and then take key points that
- 22:42we can identify in both the photo and on the
- 22:46model that are arrayed both in the XY and the
- 22:51Z. planes. So you want some things that are close
- 22:56to the ground, but you don't want all the things
- 22:58to be close to the ground. You want some of them
- 22:59to be raised in space to create the proper alignment
- 23:04for the skid loader and our reconstruction. Similarly,
- 23:10we did the same thing with the truck and we could
- 23:12define the space then that existed between the
- 23:16truck where it was parked and the skid loader
- 23:19where it was parked. And that's using features
- 23:21from the photographs of the scene. Now we know
- 23:24the space that the concrete buggy was being occupied
- 23:29in, and we could begin to look at the injuries
- 23:32that our victim suffered and figure out how he
- 23:37became entangled between the concrete buggy and
- 23:39the truck. And in fact, the digital analysis
- 23:44allowed us to go back into the lab and build
- 23:47a physical scenario to sort of be able to interact
- 23:51with a real surrogate with the back of the truck
- 23:54and the concrete buggy to start to explore, how
- 23:58could this happen? How did this person get trapped
- 24:00between the concrete buggy and the truck? And
- 24:03we learned that what had happened was, one, the
- 24:10space that was being operated in was a little
- 24:12too tight really for operating the buggy. The
- 24:16operator had successfully maneuvered it in this
- 24:18space a few times, and it... could be successfully
- 24:21maneuvered, but the placards on the machine suggested
- 24:26that it should not be operated in such a tight
- 24:28space. And the other thing we learned that was
- 24:31interesting was that once the operator got pinched
- 24:37between the buggy and the back of the truck,
- 24:41their torso pushed on part of the mechanism that
- 24:44allowed them to release the drive system on it.
- 24:47And so the machine kept pressing them into the
- 24:51back of the truck, and they weren't able to stop
- 24:54that in a timely fashion to avoid injury. This
- 24:59view gives us a variety of those depictions.
- 25:01We could very accurately indicate where the truck
- 25:04and the skid loader had been located, even though
- 25:06we weren't there at the time of the accident.
- 25:08And those measurements weren't taken at the time
- 25:11of the accident. We could reconstruct it because
- 25:13we had this 3D scan that we could pull it all
- 25:16together with. We could show how one could maneuver
- 25:21the buggy, and you could tell from the previous
- 25:24images we figured out sort of what happened in
- 25:28this particular accident. This led to an early
- 25:31settlement in this case. The visuals really had
- 25:36a way, a powerful way, I think, of shifting the
- 25:39conversation to... what would have perhaps happened
- 25:44to a pretty discreet and narrow set of possibilities
- 25:48in terms of what happened. And, you know, in
- 25:52many of these cases, this is kind of my personal
- 25:54reflection on these things, but you look at a
- 25:57lot of these cases, and it does make you wonder,
- 26:00like, man, if we're thinking safety first, why
- 26:02didn't someone just move the dump truck ahead
- 26:05a little bit? You know, maybe there was a car
- 26:07parked on the other side of it. I don't... know
- 26:11why, but it would have been relatively easy to
- 26:15create more space in this case. I mean, they
- 26:17were lifting this stuff up there in a skid loader,
- 26:19so you have to drive it five more feet. It probably
- 26:22would have changed this scenario. And I think
- 26:26these visuals are a powerful way, even for someone
- 26:30who may not be trained or have ever operated
- 26:32in a space like this, to imagine how something
- 26:35like this might have happened and what could
- 26:36have been done to prevent it. So capturing time
- 26:41is the next topic. The previous one you noticed,
- 26:46there was no moving animation in that case. We
- 26:50literally used these 3D animation tools to let
- 26:55space reveal what likely happened in this case.
- 27:01The challenge about actually animating something
- 27:03is you need to now layer on the element of time.
- 27:07And what might be going through your head is,
- 27:10where do you get that? How do you understand
- 27:13time? Well, this is something we've all seen
- 27:17a lot of, not just a knucklehead mowing his lawn,
- 27:21because that's me just... mowing my lawn here
- 27:24in Atlanta. But this is ring camera footage.
- 27:27So, you know, like it or not, my ring camera
- 27:29captured me mowing my front lawn. I didn't really
- 27:32know it at the time. But when I was preparing
- 27:33this presentation, like, gee, I need some security
- 27:36camera footage I can share and like, well, here
- 27:38I am mowing my lawn. So this is how we can capture
- 27:42time. You can see very precisely where I'm moving.
- 27:46The video has a time scale on it. But there are
- 27:49a few things that you should be aware of. when
- 27:54you're thinking about using security camera footage.
- 27:57And they relate to how companies make this footage
- 28:01small enough to store it. So one is that the
- 28:06frame rates of these videos may not be consistent.
- 28:10So if you imagine 10 frames a second as... a
- 28:17particular video, I'm just using that as a round
- 28:18number, you know then how things are moving because
- 28:22you know in every frame it's a tenth of a second.
- 28:25And so now you have this nice linear time scale
- 28:28associated with what you're seeing in the video.
- 28:31So if I wanted to measure how far I walked in
- 28:34a given period of time, if I knew it was 10 frames
- 28:37a second, that's a really easy calculation to
- 28:40make. However, Most security camera footage or
- 28:43much security camera footage is has a variable
- 28:46frame rate because they don't really want to
- 28:49store a lot of frames where things aren't moving
- 28:51and so you should be aware of that because if
- 28:56Someone who is working on this a technologist
- 28:58working on this has the original video They will
- 29:02they can tell exactly what the frame rate is
- 29:05over any given second of the video If we have
- 29:09a video that is an export or is maybe recorded
- 29:13with a cell phone off the screen of something
- 29:16like this screen, we have no way to know what
- 29:19the original frame rate was in the video. And
- 29:21unless we see something that looks really anomalous,
- 29:25we may be mistaken in our interpretation of time.
- 29:28The other important information here is that
- 29:31this is a highly distorted image. So if you haven't
- 29:35looked like this pillar is curved and you can
- 29:39see the rain spout on my house is like curving
- 29:42very extremely. can tell you with confidence
- 29:45this pillar is not curved on my house. Similarly,
- 29:49at this side of the frame, you can see the exterior
- 29:51of the house here is bending. That's just a function
- 29:55of a camera trying to capture a really wide field
- 29:58of view. That can be corrected, but again, if
- 30:01we have the original footage, we know those settings.
- 30:05There's metadata associated with the frames that
- 30:07will help us correct it very easily. If we don't
- 30:10have the original footage, it's more difficult
- 30:14to correct and maybe less precise. How else can
- 30:19we capture time? Well, there are lots of things
- 30:23that happen that we can't see. And so how do
- 30:26we capture time for something we can't see? Well,
- 30:29this is an example of using simulation to capture
- 30:32something we can't see. So we might see the movement
- 30:35of this vehicle during an accident sequence.
- 30:38It might have been captured on a security camera.
- 30:40So we know precisely how the vehicle moved, but
- 30:43we can't see the occupants in the security camera
- 30:46footage. And perhaps there's... injuries that
- 30:50are allegations of something that went wrong
- 30:52with a vehicle or something. Well, we can use
- 30:55tools like simulation tools, this particular
- 30:57one is Matimo, that allow us to simulate the
- 31:00occupant's motion in the vehicle and may give
- 31:03us some sense through an expert's interpretation
- 31:06of what the movement was of the occupants during
- 31:09an accident sequence. So again, I've kind of
- 31:14gone on about some pretty technical things. Any
- 31:17questions so far, Suzanne? No questions, just
- 31:22yet. OK. Cool. Well, we are going to switch back
- 31:27into case study mode. And I want to tell you
- 31:30about a case where we sort of put this all together,
- 31:34right? We're going to collect our x, y, z data.
- 31:38And we're going to put time into play. There's
- 31:43another element that I'm going to layer into
- 31:45this one that's also particularly fun because
- 31:48it has a lot to do with this particular case
- 31:52and why it resolved early. And that is a big
- 31:57word called conspicuity. I love saying it because
- 32:00it makes me sound smart, but it really means
- 32:02what you could see. Like, what was someone able
- 32:06to see in this particular circumstance? And it's
- 32:10critical in this case. So I mentioned a couple
- 32:14caveats to security camera footage. Frame rate,
- 32:19lens distortion. Another caveat is many security
- 32:24camera cameras are designed to see better than
- 32:28human beings can see. They have the ability to
- 32:31see better in low light conditions. They can
- 32:34see better, so they can see scenes better at
- 32:37night and things like that. And so when we evaluate
- 32:41this information, we can't always say, oh, well,
- 32:44you can see this in the security camera footage,
- 32:46so this person should have easily seen that.
- 32:49You have to consider what human perception would
- 32:51be like. And there's a very rigorous way to analyze
- 32:54a scene and its lighting with special photography
- 32:58and special targets that are used to determine
- 33:03what the human eye could see in a given scenario.
- 33:07And I just want to give you a heads up. That's
- 33:09embedded in this case study, and we can talk
- 33:11about that a little bit, too, and kind of how
- 33:14reliable that is. So in this particular case,
- 33:18it involves a bus that struck a pedestrian at
- 33:23an intersection. This image is actually partway
- 33:27through the animation. It shows the bus's approach.
- 33:31to this particular intersection, then I'll just
- 33:34kind of share an interesting anecdote about this
- 33:37because I live in Atlanta now, and this accident
- 33:41actually happened in Atlanta. But at the time
- 33:46this accident happened and the time we were doing
- 33:48all the analysis of the data, I lived in Ames,
- 33:52Iowa. And that's where our visualization team
- 33:55is located, where the majority of that kind of
- 33:58work happens. And this particular case was being
- 34:01worked on. So I had sort of been virtually in
- 34:06this intersection many, many times as this animation
- 34:10was developed. And we were consuming the data
- 34:13and putting it all together. This is... You know,
- 34:18this case resolved years before I moved to Atlanta.
- 34:21I moved five years ago, and I'm a cyclist. I
- 34:24love biking around the city. And one Saturday,
- 34:28I was out on a ride, and I rode into this intersection
- 34:31totally randomly unbeknownst to me. And it was
- 34:35the weirdest experience I think I've ever had
- 34:38in an interesting way, because it was this...
- 34:40deja vu that couldn't have been like, I've never
- 34:43been here before. I've never been on this side
- 34:45of the town. And yet I'm riding into this place
- 34:48that I know. It was also really satisfying because
- 34:51it was like, wow, this, when I figured out why
- 34:54I was having this deja vu, it was like, you know,
- 34:57like I look, I know this data is really reliable,
- 35:00but wow, it's like really reliable in this interesting
- 35:04human tangible way. And so this one kind of has
- 35:08this. personal context to it that many of these
- 35:12end up having when we work on them in this way.
- 35:16I digress. What do we do first? Well, the first
- 35:21thing we want to do is capture space. It's really
- 35:24important that we understand the space at a high
- 35:28level of detail. These are like the refinery
- 35:33I showed you. the outcome of a series of laser
- 35:37scans, where we position laser scanners up and
- 35:40down these streets in both directions to capture
- 35:44information that we needed to understand the
- 35:46bus's approach to the intersection, things beyond
- 35:49the intersection, and things in either direction
- 35:52to the sides. For your orientation, the bus was
- 35:56actually coming from really the top center of
- 35:59this screen down through to the bottom. left
- 36:04hand side of this screen. So it was passing down
- 36:08this Ralph David Abernathy is the name of the
- 36:12street. The intersecting street is Lee Street.
- 36:17This is now looking again back up the street.
- 36:21So the bus would be coming toward us. um in this
- 36:25particular view and again this is our view of
- 36:27the intersection one of the things that was really
- 36:29crucial in this case because we knew it was going
- 36:32to be you know that nighttime visibility was
- 36:34going to be really important in this case was
- 36:36that our scan captured all the light sources
- 36:39in this scene it was nighttime and so much of
- 36:44the illumination in this intersection was going
- 36:46to be street lights and you know storefront windows
- 36:50and things like that. So we needed to capture
- 36:53all the light sources in the scan so we could
- 36:56very accurately place light sources in those
- 36:59fixtures. This is just a little closer look at
- 37:04the intersection the crosswalk that the pedestrian
- 37:06is going to walk on in this particular screen
- 37:09is Going to go from the right side of the screen
- 37:12to the left side of the screen across You can
- 37:14see the whole crosswalk there and again the bus
- 37:16is coming toward us in our direction And then
- 37:20finally, in this view of the scan, you're looking
- 37:23down the street kind of from the bus driver's
- 37:25perspective. This is what the approach to that
- 37:28intersection would have looked like from the
- 37:31perspective of the bus. And the elevated thing
- 37:34that you see in the distance is actually a railroad.
- 37:37The MARTA line actually runs across on the first
- 37:41bridge, and there's a freight train line that
- 37:43runs across a second bridge that crosses the
- 37:45road. I mentioned before, sometimes we source
- 37:51things from other things than laser scans. And
- 37:53in this particular case, we often use a hybrid
- 37:56approach where we scan the bus, but then we buy
- 37:59a model that might be commercially available
- 38:01that is that bus, and we just make sure that
- 38:04it's dimensionally accurate and that everything's
- 38:07positioned the way that it was on the bus. um
- 38:10that we have access to because you can imagine
- 38:13with a visibility case we're going to be really
- 38:16worried about anything like side view mirrors
- 38:19things that might have blocked the driver's point
- 38:22of view we want to make sure we have those positioned
- 38:24in in the 3d bus exactly as they were on the
- 38:28bus that was involved in the accident similarly
- 38:31There are human models, a variety of sources,
- 38:36and there are tools that allow us to build those
- 38:39models in a scalable fashion. So once we have
- 38:41information about folks involved in the accident,
- 38:45to the extent we have information on them, we
- 38:47can actually build a dimensionally accurate model.
- 38:51In the case of conspicuity, we want to also make
- 38:54sure clothing, if there's any reflective properties
- 38:57to the clothing, colors of clothing, things like
- 39:01that, that we have that right. So again, security
- 39:06camera footage is going to be really important
- 39:08to us. And one of the things about city buses
- 39:11is they are bristling with security cameras or
- 39:14buses are... cameras all over them. And so we
- 39:17have, in this case, four of the cameras. There
- 39:20were at least nine, I can tell from the layout
- 39:24here. This is the forward -looking camera on
- 39:28the bus. The one to the right is actually a camera
- 39:32that's focused on the bus driver. The lower camera
- 39:37is looking at the front entry door, and then
- 39:40there's a camera on the outside of the bus looking
- 39:42back down the side of the bus along the side.
- 39:46I'm not going to show the video. It's a little
- 39:48bit graphic, but it but what this gives us is
- 39:52a record of the bus moving down the street. And
- 39:55initially there were witnesses that said the
- 39:57bus was speeding. It was driving erratically.
- 40:01There were a number of allegations about the
- 40:02bus's movement that was. actually really straightforward
- 40:06to pin down with this video. So we knew exactly
- 40:09how fast the bus was moving because we could
- 40:12move it frame by frame through the 3D scene using
- 40:15this video footage. We could tell the path of
- 40:18the bus. In fact, we were even able to track
- 40:21the driver's behavior because of the camera that
- 40:24was focused on the driver. So I'm going to show
- 40:28you this is a top down view. So this would be
- 40:31a typical orienting animation that we would use
- 40:35to help viewers just begin to digest the scene,
- 40:40like what happened that night on Ralph David
- 40:43Abernathy. So a couple things that are additive
- 40:47here that you should understand. So the bus is
- 40:49on the left side of the screen here. Before I
- 40:52put this in motion, I want you to notice that
- 40:54there are the two triangles coming out of the
- 40:56front of the bus that are kind of weird. those
- 40:59are depicting the driver's field of view. And
- 41:03so we have... high acuity of vision when it comes
- 41:08to our foveas, a specific part of our retina
- 41:11that's very sensitive. So this is showing in
- 41:15red the field of view that would have been high
- 41:18acuity for the driver. And then the blue triangle
- 41:21is showing the field of view, which we would
- 41:23call parafovial. It has less acuity than the
- 41:28foveal view. The other thing you're going to
- 41:30notice in the readout is at the top there's a
- 41:33time to impact and then the speed of the bus.
- 41:36The speed limit on this chunk of roadway, if
- 41:40I'm recalling rightly, I think is 35 miles an
- 41:42hour. The bus was never even really close to
- 41:45the posted speed limit as it approached this
- 41:48intersection. So I'm going to put this in motion.
- 41:51Oh, the other thing I should mention before I
- 41:53do that is this the orange or the yellow circles
- 41:56there on here are circling pedestrians. There
- 41:59were other pedestrians out there, but from the
- 42:02experts analysis of the head movements of the
- 42:05driver, these three were particularly important
- 42:08to the driver's attention. And it's the pedestrian
- 42:11that's in the crosswalk that will be struck.
- 42:15So here we put the bus in motion. We're seeing
- 42:20where the driver's attention is you see at the
- 42:23very last second the driver's attention Focused
- 42:28on the pedestrian that he struck and you kind
- 42:30of wonder why because that pedestrian was really
- 42:34the one that was Really out in front of the bus,
- 42:37right? Why was so much attention focused on this
- 42:42other pedestrian? Well When you approach this
- 42:46intersection, it's important to understand from
- 42:47the driver's point of view that the lighting
- 42:50is really interesting. There's oncoming traffic
- 42:54at this particular intersection. And the way
- 42:57the street is laid out, the oncoming traffic
- 43:00is actually kind of coming up a slope to the
- 43:03stoplight. So the headlights are particularly
- 43:05blinding because they're really pointing more
- 43:09toward your eyes. In this animation, we're going
- 43:12to show the same sequence from the bus driver's
- 43:17point of view. So now the triangles are circles.
- 43:20And so the red circle represents the high acuity
- 43:24vision, and the blue circle represents the lower
- 43:27acuity vision. And we're going to be able to
- 43:29see where the driver's attention was focused
- 43:32as he approached the intersection. The other
- 43:36thing I want you to pay attention to simultaneously
- 43:40is the visibility of the pedestrian that's in
- 43:43the crosswalk. I'm going to give you the clue.
- 43:46You're going to know she's there, but imagine
- 43:49not knowing she's there and where your attention
- 43:51might be as you approach this intersection and
- 43:54note that the light is green. The pedestrians
- 43:57do not have the right of way in this particular
- 43:59case. So that's the sequence from the bus driver's
- 44:13point of view. And I think oftentimes the power
- 44:17of 3D animation and this type of visualization
- 44:21is not so much the diagrammatic interpretation
- 44:28like the first one I showed you that was top
- 44:31down. That's really helpful in getting viewers
- 44:35aligned with what we're talking about and what
- 44:37really happened in this scene. But when you can
- 44:41then take all that data, we know how the bus
- 44:43moved, we know where the driver's head was moving,
- 44:46we know the lighting for the scene, we know that
- 44:49the geometry of it is precise, and we put you
- 44:53in that driver's seat and you can experience
- 44:55from that driver's point of view, it is really
- 45:00influential when it comes to resolving matters
- 45:04like this one. Any questions so far Susanna we
- 45:14so far so good or interesting questions popping
- 45:17up. I just had one question pop up actually Alexander
- 45:21Ginsburg asks how much approximately did the
- 45:253d animation cost in the bus case? Hmm. So in
- 45:29the bus case that animate the total cost of creating
- 45:33that animation as I recall was around $65 ,000
- 45:38I think was spent by that client And I, I throw
- 45:42that big number out there, but it doesn't happen
- 45:47that way. So, and the reason I said that is it
- 45:52happens in stages and the team, the legal team
- 45:57makes a lot of choices along the way. So the
- 45:59first thing we do and we were called into a case
- 46:02like this and we get it, we sort of begin to
- 46:04understand it is very likely we're going to go
- 46:06out and do a laser scan. When we do that laser
- 46:09scan, what I showed you a few images, what we
- 46:12would typically do is sit down with the client
- 46:14and look at the laser scan and go through all
- 46:17that data and kind of figure out like kind of
- 46:20maybe we even get in VR and we kind of come down
- 46:22the street from the perspective of the bus and
- 46:25we start to see what's going on. We know it was
- 46:28nighttime. We maybe have photographs from the
- 46:30scene. You start to piece together what's going
- 46:33to be important. That Phase of it is less than
- 46:38$10 ,000 to do that part. It depends on the complexity
- 46:42of the scene. It depends on a few things. But
- 46:45it can be anywhere from $3 ,000 to $10 ,000 to
- 46:49get that view. And oftentimes, it's at that stage
- 46:53where we're deciding, should we move ahead or
- 46:57is this something where we want to go in a different
- 47:00direction? And then the next step would be to
- 47:03start layering in. Well, let's let's get a scan
- 47:05of the bus. Maybe that's really important to
- 47:07look at. Maybe there's something abstract. Maybe
- 47:09the windshield was damaged on the bus or there's
- 47:11something about that. So we begin to look at
- 47:15different questions pieces at a time. So it ends
- 47:17up being a series of bites at removing uncertainty
- 47:23and knowing more that ultimately gets us to a
- 47:27visual like this. And usually once we're at the
- 47:29point of you know, going down the road of really
- 47:32rendering a visual that we're going to put out
- 47:34in front of maybe the opposing council or for
- 47:37a mediation or that kind of thing. We have pretty
- 47:41good sense of, you know, what our story looks
- 47:45like in the case. So I hope that's responsive.
- 47:50That's always kind of a it's an interesting question
- 47:52because I think there's sometimes sticker shock
- 47:53at the price, but it does it sort of happens
- 47:56over time. And that's a really important factor.
- 48:03Any other questions, Suzanne? Not seeing any
- 48:08other questions just yet. OK, well, I'm going
- 48:11to do another case study where we layer in another
- 48:14piece of knowing. And in the previous one with
- 48:18the bus, you might have noticed that the pedestrians
- 48:21were like. a little bit robotic. Maybe they weren't
- 48:24moving as smoothly as others. And that case is
- 48:29from a long time ago when we didn't have the
- 48:32technology we have at our fingertips now. And
- 48:34one of the really interesting technologies, really
- 48:38important technologies that we leverage when
- 48:41it comes to understanding how humans interact
- 48:43with the environment as human motion capture.
- 48:46And this has been used in the movie industry
- 48:48for a long time. It's sort of needed to reach
- 48:52I need to cover a couple things. It had to get
- 48:54to a place where we could do it out in the real
- 48:57world for it to be particularly valuable for
- 49:00us. And it had to get to a space where it was
- 49:03also affordable enough that we could deploy it
- 49:06in cases where it scaled properly. So that happened.
- 49:12We now have a human performance lab in our office
- 49:15in Ann Arbor, Michigan. But we also have Motion
- 49:20capture suits that can go anywhere and so I'm
- 49:23going to show you this particular case now I'm
- 49:24going to warn you that in the in the security
- 49:27camera footage I'm going to show you you are
- 49:30going to see a runner get struck by a vehicle.
- 49:33I will tell you she survived She had injuries
- 49:37If that's if that's troubling to you, just don't
- 49:42watch for a little bit here, I will tell you
- 49:45when that has passed. It's not like she breaks
- 49:47into a million pieces or anything, but it is
- 49:50kind of violent. So this is the point where you
- 49:55might want to look away and tee this up. This
- 49:57is security camera footage that's on a building
- 50:00that houses a gym. And this particular footage,
- 50:05I'm going to play it now. You'll see a jogger
- 50:08in the street back behind the trees. So you see
- 50:11her pass between the trees and then you see this
- 50:14vehicle approach and it strikes her, throwing
- 50:17her into the street. The vehicle was animated
- 50:21like we covered up the vehicle just to protect
- 50:23the party in this particular case. So, OK, now
- 50:28you can look back. All the violence will be only
- 50:31animated at this point. This gives you a map
- 50:37of the layout of the scene. You'll recognize
- 50:40it immediately as a laser scan. This is a scene
- 50:44like everything else I've shown you where we
- 50:46went out and immediately scanned the scene. The
- 50:49building where the camera was mounted is right
- 50:51at the bottom of the screen. And you see we've
- 50:53marked the security camera location here. The
- 50:57interesting thing about this case is these two
- 50:59trees, the tree to the left actually blocked
- 51:03our view of the actual vehicle striking the jogger.
- 51:09But it's amazing how data can get pieced together.
- 51:13And because if you watched it, you saw how the
- 51:17vehicle threw her, how she rode a little bit
- 51:20on the hood, the vehicle threw her, where she
- 51:22landed. All of that is dictated by the laws of
- 51:26physics. And so the throw distance and a number
- 51:30of these things help us map where things happened
- 51:35and with how much force and how fast things were
- 51:38going. So all of these little pieces in that
- 51:41security camera footage are clues that help us
- 51:44piece this back together. We've got the scene.
- 51:49We can also, because we have the 3D scene, we
- 51:53can accurately map the vehicles and the pedestrians
- 51:57back onto the scene in 3D. So if you imagine
- 52:02we take a scale model of the vehicle, a scale
- 52:04model of the pedestrian, and by matching frame
- 52:07by frame, we can figure out what their path was
- 52:10on the roadway, even though that's kind of hard
- 52:13to discern when you're just looking at the video
- 52:15itself. When you take it all apart, and then
- 52:19put it back into the 3D model, these locations
- 52:22can be remapped. To lay out the scenario for
- 52:26you so you understand the sequence a little better,
- 52:29at this gym, part of what some of the folks were
- 52:32doing was a timed run, and they would go out
- 52:35from the gym and they would run two laps where
- 52:39they would run up this roadway to somewhere in
- 52:42this neighborhood. They would come back. You
- 52:45can see the jogger's path here. coming up this
- 52:49way, when they reached this mailbox, they were
- 52:52supposed to turn left, cross this crack in the
- 52:57driveway, and then they would either go out for
- 52:59a second lap, or if it was their second lap,
- 53:01that was their time when they crossed this. So
- 53:04she's coming back along this roadway, getting
- 53:07ready to turn into this driveway, totally unbeknownst
- 53:11to us or anybody until we talked to people at
- 53:13the gym. So this is a view. This is also this
- 53:19is a drone photo of the of the accident scene
- 53:23or the site where the accident happened. So this
- 53:26gives us an overhead view. You can see the security
- 53:28camera footage on the building here. And then
- 53:31the picture on the right is the laser scan of
- 53:34this particular scene. So we know a lot about
- 53:40this scene, but what we really wanted to do is
- 53:43really capture some accurate human motion. So
- 53:47we employed a physical surrogate. This is a person
- 53:51who's about the same size as the runner that
- 53:53was struck that day. And this is a video that
- 53:56I'm going to show you of a motion capture run.
- 53:59So she is wearing a Lycra suit that has 17 accelerometers
- 54:04embedded in it. And they're embedded in the various
- 54:07segments of her body. So it's actually measuring
- 54:12the movements of 17 different locations on her
- 54:18body. So what we did was painted markers on the
- 54:23ground that we could tell from the video footage
- 54:26where the runner's feet landed. And we asked
- 54:29this runner to run back over the footsteps of
- 54:32the accident runner. This is what that data looks
- 54:36like. This is just the output from the software
- 54:40itself. So there you can see her run now digitally.
- 54:44Each of those kind of chunks of her body represents
- 54:47the segments that are measured by the accelerometers.
- 54:51Cool thing about this data, I'm going to show
- 54:54you how it informs animation, but... It also
- 54:58is capturing forces, accelerations, all kinds
- 55:01of things on every one of the 17 segments of
- 55:04our body. So a biomechanics expert can go in,
- 55:09we can reenact things and ask, gee, what were
- 55:11the forces on this person's thigh bone when they
- 55:16were doing this particular activity? So now I
- 55:20want to show you. Keep in mind, we did not animate
- 55:24the runner's motion based upon the runner in
- 55:27the video. We animated the runner's motion based
- 55:30upon the simulation that we did with our surrogate
- 55:34runner. So here's a frame of the securities camera
- 55:37footage where you can see the runner that was
- 55:40struck. And here's a frame of the animation with
- 55:43our surrogate runner passing through that same
- 55:46space. Now here's the animated frames. And what
- 55:52I'm going to show you now are the animations
- 55:54that were used to settle this particular case.
- 55:57I will tell you now, you're not going to see
- 56:00the throw. So this is less violent. The impact
- 56:03is going to happen behind this tree. And, you
- 56:07know, we don't really care about what happened
- 56:10after the impact in terms of the animation. It
- 56:13was useful in figuring some things out about
- 56:15the dynamics of this accident. So this is not
- 56:18going to appear as violent as the video did.
- 56:22So this is the animation Our runner is going
- 56:26to pop into the frame here between the trees
- 56:30and then here comes our truck and The accident
- 56:34occurs truck magically rolls through without
- 56:36the runner This viewpoint is from the view of
- 56:40the security camera footage and it's got a purpose.
- 56:44It's, I think, really important to show the viewers
- 56:47that we have accurately reconstructed what happened
- 56:51in the security camera footage in this 3D environment.
- 56:55So this is sort of maybe a confidence builder.
- 56:59It really lends integrity. to all the work that's
- 57:03gone into this without having to just you know
- 57:06brag about all the work that went into this.
- 57:09This is now our top down view like you saw in
- 57:11the bus scene and you can see the vehicle approaching
- 57:15and then all of a sudden the runner makes this
- 57:18hard left turn at the mailbox in front of the
- 57:21vehicle. Another point of view again because
- 57:26we've got this accurate depiction now We can
- 57:29show as many points of view from any angle that
- 57:32you want to see it There's the strike and then
- 57:37again kind of like the bus I think the one that
- 57:39is really important is the driver's view from
- 57:41the vehicle So she's running along the shoulder
- 57:44the driver starts to move over to pass and then
- 57:47she makes us hard left right in front of the
- 57:50vehicle Our experts that analyzed this from beginning
- 57:55to end, it concluded that the driver really didn't
- 57:59have an opportunity to avoid this particular
- 58:01accident. That when the first clue sort of came
- 58:04that the runner was going to move into the path
- 58:07of the vehicle, when you look at reaction times
- 58:10and things, there was just not enough time to
- 58:13respond to the runner. for the last two minutes
- 58:23here about virtual reality. I'm gonna just compress
- 58:25it a little bit because I mentioned earlier how
- 58:28we use it in this context, like early on, looking
- 58:32at data, getting the entire team aligned around
- 58:35data that we've collected and what maybe we should
- 58:39do next, like both from the logical standpoint,
- 58:44like what should we do, but. You know, the question
- 58:47was asked earlier about budget from a budgetary
- 58:50standpoint. If we're limited in terms of a budget
- 58:52going forward, what's the most important thing
- 58:54we can do next given the budget? And so I think
- 58:57that's really important to entertain as well.
- 59:00So early on is important, but we also use it
- 59:03late in the game, especially in cases like this.
- 59:06This is a mid -air collision that happened between
- 59:09a helicopter and an airplane at an airport on
- 59:12the East Coast. In many of these cases, we know
- 59:16a lot. We know where the aircraft moved, but
- 59:19we don't necessarily know where the crew was
- 59:21looking. They usually don't survive these things.
- 59:24And so what we can do is put professional pilots
- 59:27in the seats of these aircraft and ask them where
- 59:33they should have. or would have potentially been
- 59:36looking in the course of an accident like this.
- 59:39In this particular case, I'm looking at the oncoming
- 59:41aircraft. There's a variety of reasons to believe
- 59:44in this analysis. We can understand why they
- 59:48possibly overlooked this particular aircraft.
- 59:50But in this case, we can pipe in the air traffic
- 59:52control. audio, any kind of cockpit conversations
- 59:57if there was a cockpit voice recorder, and time
- 59:59that all up, which is super valuable for an expert
- 1:00:03that's going to testify on piloting issues. Finally,
- 1:00:07this is a pedestrian that was struck by a train.
- 1:00:10Again, we often use VR to try to evaluate what
- 1:00:14information did this pedestrian have access to
- 1:00:17once we reconstruct all this. This person obviously
- 1:00:21didn't see this train coming and was struck by
- 1:00:23it. They thought that the alarm was going off
- 1:00:26for the other train. But this gives us real human
- 1:00:30insights. Instead of spreadsheet after spreadsheet
- 1:00:35of data, we can take people in and have real
- 1:00:39human insights on these kind of events. So I
- 1:00:43hope this was helpful. At minimum, I hope it
- 1:00:46was interesting. And I'm happy to hang around
- 1:00:49for a while and entertain questions if you have
- 1:00:52them. Right. Sorry. Excuse me, Dr. Fox. It looks
- 1:01:00like we don't have any questions at the moment.
- 1:01:03Okay. Well, thank you so much for your attention.
- 1:01:07This has really been fun. This concludes today's
- 1:01:11presentation. Thank you so much, Chuck. Dr. Fox
- 1:01:15for your time today and thank you so much to
- 1:01:17our audience. Have a great day everybody. Thank
- 1:01:20you.