← Reshaping Workflows with Dell Pro Precision and NVIDIA RTX PRO GPUs
Reshaping Workflows with Dell Pro Precision and NVIDIA RTX PRO GPUs
Bringing Geospatial Data to Life with Esri and Dell
00:43:14
Discover how Esri, Dell, and NVIDIA bring geospatial data to life with immersive 3D, digital twins, and fast, AI-powered decision tools.
Loading summary
Transcript56 lines
- [00:00]
A
Foreign. Welcome to Reshaping Workflows with Dell Pro Precision and Nvidia, where innovation meets real world impact in high performance computing.
- [00:20]
B
Welcome back to another exciting episode of Reshaping Workflows with Dell Pro Max and Nvidia RTX Pro Pro GPUs. I'm your host, Logan Mahler. So today we're doing a couple of cool things. One, we're talking about a, an ISV which we haven't really talked a whole lot about. We've had partners, we've had AI companies, we've had everything kind of in between. Today we're going to be talking about Esri that's known for their ArcGIS product that basically kind of puts together space and time. I know that sounds completely crazy, but as we get into it, you're going to absolutely understand exactly what I'm talking about. So with this we have both Rex with us from ESRI as well as Pat. Let's get started with introductions. Hey Pat, if you don't mind, take 30 seconds, introduce yourself what you do at ESRI a little bit on your background so all the listeners kind of have an understanding where you're coming from today. So with that Pat, go ahead.
- [01:13]
A
All right, thanks Logan. I'm Pat Wallace, I lead esri's Innovation lab. What do we do there? Really looking at gap fitting new capabilities into ArcGIS by looking at end to end workflows for our users and if there's not a neat fit in our off the shelf products. We work extensively with our customers, our partners, internal stakeholders like different product teams like the product team. Rex leads to draft out scopes for these research projects and then we iterate through to show value and then know after we show value working with our stakeholders, we then work shoulder to shoulder to onboard, you know, our findings into our tech stack. So the whole, whole goal is to make better software for our customers.
- [02:02]
B
Love that. If you weren't making better software you probably wouldn't be in business. So I love that. Makes total sense. All right, Rex, what about you?
- [02:09]
C
Great, thanks Logan. Rex Hansen, I'm the product manager for our native and our game engine SDKs, our ArcGIS maps SDKs for native and game engine development. We have a team of over 100 engineers that are responsible for building out developer tools to light up really cross platform solutions targeting mobile desktop, embedded situations or devices. We work with a host of customers and partners across industries to build out focused applications that deliver mapping for visualization, exploration for real time situational awareness for analytics and simulation workflows. Some of the more exciting things that we're doing here lately is basically combining the power of GIS with the capabilities that game engines provide. And so that's something that my team is investing time and effort and research into in trying to make the most of really basically bringing your geospatial data to life with game engine capabilities.
- [03:04]
B
I love that. So, got a lot of questions and I'll be the first to admit we were supposed to have Ken here and let's just say our lovely host lost a bet. That's neither here or there. That's going to be a fun little tag on later. But without Ken here, I don't know much about esri. So this is going to be great. This is going to be a great educational episode, not only for me, but for everyone out there. So when you say geospatial and either one of you can answer the question, whatever works. When you say geospatial, are we talking about like Gaussian splatting? Are we talking about like Google Maps, kind of like GPS data? Define what geospatial is first.
- [03:38]
A
We'll start with the fact that Esri builds ArcGIS, it's the world's leading GIS mapping software. You know, that's the geospatial aspect to it. Right. So what is gis? And this is the mapping and data analysis tech that's powering business decisions for our customers across government, you know, private corporations, non governmental organizations, you know, it's the science aware that's quietly changing the world around us. It's kind of an all the above because it's a system approach, right? We're providing capabilities that fit the people, the process, the data and then the tech that pulls all these bits together. Does that help?
- [04:21]
B
Yeah, it makes perfect sense. Rex, I think you're going to say something maybe, or not.
- [04:24]
C
If you think about it like we are, we are the experts in managing space and time, right? And every industry is affected by space and time in some capacity. And so we have, I think over 700,000 organizations that we work with globally to be, you know, to be able to champion and deliver a solution that allows them to manage, enables them to manage space and time effectively. To be the world leader in GIS
- [04:47]
B
and GIS space and time. I mean, that's pretty easy. I mean, have you guys trademarked that? Because you should, because they're pretty smart.
- [04:55]
A
It's afternoon, Logan.
- [04:56]
B
Yeah, or top we or hi.
- [04:58]
A
Getting on the lawyers.
- [04:59]
B
Now in terms of, you know, GIs, you mentioned kind of SDKs that develop off of that we'll come back to that in a second. But kind of ArcGIS is it a, I mean, sounds like kind of an ingestion platform where you're providing the intelligence within this geospatial to ultimately feed into other things, was what it sounds like. So give me kind of an example of what can be fed into ArcGIS. I mean, I know there's different industries, there's a lot of different data, but like what is the most common things that are fed into your platform?
- [05:28]
A
Digital twin for me, I think about marriage of digital transformation and decision support. You know, kind of connecting the loop on the OODA loop right through technology to help our users make better decisions in the world and keep that connected correspondence between the world and the digital representation of it. So to do that we can bring in all sorts of data, you know, tabular data, two dimensional data points, lines, polygons, three dimensional data that represent things in the world, concepts in the world, like political boundaries, you know, and then marry that to different business systems, allow our users to create maps that are relevant. You know, if there's a need for a human user to see a visualization. But it could be behind the scenes, like you alluded to, Logan. It could just be like, I just want to use the spatial bits and analysis to answer some natural language questions. We do all of the above, right? So we, we provide outputs via services. We provide outputs that are information products, that are maps, interactive web maps or maps on, you know, leveraging our, our mobile SDKs, right, and the ability to, to print it out because people still need printed maps, right, Using this technology.
- [06:46]
C
That's true. I was going to say, I guess I was going to add to it a little bit there, Logan. You know, the topic or the term metaverse was floated around a few years ago and it probably got overused in a variety of ways. But if we think about Metaverse being basically a digital twin of the universe, starting with the Earth, I think that's, that's our perspective. But it's not necessarily just a digital twin. It's a digital twin of reality, actually starting with the Earth, to make our physical experience, our physical world a better place. Right? So it's not just something to escape to, something to simulate against, something to, you know, to exist within. It's more like basically it's supplementary, right? It's designed to be able to enable industries, enable businesses, enable organizations to be more successful in managing their physical reality, physical space. And so while we do see folks, if you think about it, you know, maps are really abstractions of reality. Right. And so whether that's 2D, where someone has to basically kind of understand what they're looking at in two dimensions and abstract that to reality, which is 3D or 4D at this stage, if we can reduce that level of abstraction so the consumers can look at a digital representation and know where they are, what they're looking at, and be able to make decisions faster, quicker, more effectively, then we can expedite, you know, some of those, those workflows. And so that's what we're doing right now. The digital twin concept is really about trying to provide a digital representation of the physical world with position and scale in place so that our customers, our partners, can make better decisions. And oftentimes that involves not just visualization and exploration, but involves things like asset management, involves things like simulation workflows and reducing cost, increasing safety. Right. So across the board. And so being able to do this effectively with folks that maybe aren't necessarily interested in the latest technical capabilities, but still have a job to do. Right. If we can do that and deliver it in a way that those decision makers don't have to figure out where they are, what they're looking at when they're looking at a digital representation of the physical world, but being able to make that connection right away, they can make the decisions faster, safer, and make better decisions over time as well. And that crosses every industry, whether you're dealing with transportation, aec, defense, public safety, utilities, you name has an impact in every one of those industries. And every one of those industries has staff, you know, that needs to be able to interact with a digital representation
- [09:08]
A
of the real world.
- [09:09]
C
And so we need to be able to deliver that in a way that's useful.
- [09:12]
B
So you talk about making decisions not necessarily cheaper, but faster, quicker decision making. Right. Let's start with Pat and then we'll go to Rex. Don't divulge any customers or proprietary NDA stuff, but give me an example, give me a real world example where a customer used Esri, you know, ArcGIS to do exactly what you said is to make a decision faster using a 3D world from a kind of a 2D reconstruction. Give it, give me an example.
- [09:40]
A
We recently were doing some research with some law enforcement organizations and you know, for like critical events and, and they really needed kind of a simple, you know, what you see, what you get type of approach that Wrecks was talking about in terms of like simplifying this kind of abstraction so you understand what you're seeing right out of the gate. And we're talking about operators in the field. So you know, 2D floor plans of this particular state capital didn't make sense. Site maps of the state capitol grounds didn't make sense. So what we did is, you know, we, we leveraged, you know, consumer grade, you know, Capture devices for 360 video and drone capture data to create like comprehensive Gaussian splatting representations of those environments. Then we did the semantic, you know, chopping it up into bits that made sense for what those operators were actually trying to achieve. Right? So you could query these data, get actionable results and intelligence to, to put mass on target in the field, to continue the law enforcement metaphor there. Right. So this is, it's not your mom or dad's mapping anymore, right? We're moving into a new paradigm where you know, we can rapidly create very capable digital twin capabilities with reality capture content, run it through the geoprocessing, and gis, you know, processing pipeline. Like at esri we're all Dell on this side, right? Like, you know, we, we have to have enough graphic GPUs to put, push this processing power through to get the result that we want, you know, so it takes energy, take, takes compute power, you know, but we can push out a result that, you know, provides real results on the ground in real time.
- [11:41]
B
That's amazing. That's a great example because it makes perfect sense, right? Like, you know, which entrance do we want to come in? Like where is the fastest safety route you can actually see it without being on the ground. So it makes perfect sense. Rex, do you have, you know, a different example? Because I think this helps illuminate what it does because it is like it's, I mean like when you say I'm creating a 3D world or it seems so nebulous. Do you know what I mean? It's not, it's not a knock, it's just most people don't understand. But when you give a real world example, you're like, oh, that makes total sense, right? So I'd love to hear example Rex, if you've got one.
- [12:13]
C
Yeah, yeah, yeah, it's true. I mean, sometimes I feel like, you know, when we talk about creating 3D worlds, oftentimes it's synthetic or artificial. In this case we have a model, right? And that's the rule. That's reality, that's our physical reality. So we have something to gauge the quality of our digital twin against. And it has to be accurate at a, at a level that, you know, the industry that's using it can make Decisions effectively and useful. So I'll say we have an example in the AEC space. I know you know, Pat was talking about law enforcement. We've got a few, few in that space. But I'll move to aec, talk about some of the, the success that we're seeing there. So there's an organization named Simic. They're Australia's largest construction company. They are already an ESRI customer. They already work with and use the ArcGIS enterprise to manage their geospatial information. They have a couple of entities that work with them. One of them is named IDD Tech. They're an engineering partner with simic. They built out a product called Tubi Builder. Tubi Builder or the Tubi suite of products is built on top of a game engine, so it's built on top of Unity. And the idea there is that a couple years ago they embarked on this effort because they wanted to provide at first a desktop experience that allowed for their site managers and supervisors to be able to plan a construction site and see it through to the end, see how the construction site is managed or maintained over time. And then also as they monitor that, as they move through the construction process, establish what went right, what went wrong, establish methods or at least refinements over that process over that time. So they're already managing their geospatial information within ArcGIS. They're managing a lot of billing information through Autodesk product. And they're bringing all this together in a desktop experience that would allow for decision makers within their organization to be able to figure out how to most effectively or most efficiently organize construction projects and monitor progress. And so they introduced this a few years ago, now it's being used in production. What they needed to do is provide a real world context and work with a local coordinate system. So they needed the precision and the fidelity that came with geospatial information, needed a canvas on which to work. They needed an area in which they could, they could establish building information that changes over time, but then also being able to manage assets. And so those assets could be vehicles, could be materials where they're located around the construction site as the elevation changes, as they're digging or as they're moving or as they're trenching, whatever they happen to be doing, or if they're shutting down or cording off streets, altering transportation pathways. As this construction project moves and changes, they can see the changes in a, in a scenario that doesn't require them to really abstract between, hey, this is a static 2D map, or this is a top down experience. They can see things in what looks like, you know, sort of a real world context, a 3D context. And they had time. Time was an aspect there that they could play back as things progressed through this construction project and they can play it forward to see what their plans are. So this is interesting and it sounds kind of like eye candy, but inevitably what it did though is it allowed for folks that were organizing these construction projects to be more efficient in how they placed assets in the progress that was taking place on the ground without having to travel to that construction site. They were also able to dedicate resources, personnel more effectively because they could get a more accurate picture of what was happening on the ground. And so what they saw over time was a reduction in cost, an increase in safety. And so what they saw was an ROI. And because they saw an ROI, even a 5 for 10% change was, you know, at scale is significant. And so that process, that solution has expanded its use within CIC and has enabled them to save money and increase safety. And that's a fantastic showcase of being able to bring, reduce the abstraction, you know, leverage the, the realism that comes with building out a game engine, 3D GIS and, and asset management workflows going forward and really combine that space and time element to make it useful.
- [16:15]
B
That's an amazing use case. I mean, I love that because you're right, like I'm on the road quite a bit and travel's expensive, right? I mean going out and having the PM or whoever on site to go visually see something when they can see it in real time. You, you said something wrecks that kind of struck a core. And I, and I might be way wrong about this, but is, you know, I've used and it's not esri, right. And obviously nowhere near a product that you do, but does do some 3D reconstruction stuff that I'll do for demos just to play around of like hey, here's my house and my dog laying on the couch or whatever. Right? Very basic. Not what you all are doing, but what you said was. Or what I heard is for me to do that it takes a little bit of technical chops, like not much but a little bit. But what I heard was, hey, there's kind of the SDK side of ESRI and ArcGIS and then there's like, I'm going to call them the quote unquote knowledge worker who doesn't necessarily know how to build their 3D reconstruction isn't doing any of the backwards math or compute behind it, but Needs to be able to log into something, go in, see play, interact, but not be technical or need to have any technical skills. So is ESRI truly, like, I could give it to my, my mom to go play? I mean, maybe it's not that easy. Uh, but is, is that really the design?
- [17:31]
A
It could be. I mean, we, I mean, the way that our, our tech stack is built is we're, we gear it so that it can easily be configured and deployed for these different, you know, user types, right? So we understand that there are users that are creators, right? We understand that there are users that are consumers of the outputs. Your mom would be a consumer of these data. Then there's the kind of mix between the two, like the, the field worker, they're a consumer, but they're transactionally capturing data along the way. So you know that that's the way that we're architecting our tech stack because we know that our customers are running businesses, right? So, so we need to help them continue to do that and be able to kind of be empathetic to those, those different user types.
- [18:15]
C
That's true. And you know, like Pat said, it really depends upon the situation, the use case there. We, we do provide a set of applications really off the shelf, you know, field staff to be able to go into the field, capture data, work with their data in the back office without having to understand or know the details about what it takes to build a gis. Right? They have a job they need to do, you know, whether it's, you know, community engagement, you know, it's utility management, it's, you know, it's capturing data in the field, it's processing data in the background in the back end across multiple industries. They have a job they need to do. And so I think ultimately what we're trying to do or what we're trying to provide is off the shelf apps that enable folks to configure solutions for them. We're also providing, of course, dev tools that allow for sometimes customers, but oftentimes our partners, sometimes ESRI actually build out bespoke or focus solutions around a specific industry. And so while our dev tools are available, a lot of times the shelf app with configuration, sort of, I guess configurable properties oftentimes is sufficient for a lot of different industries to meet their workflows. And again, it's about being as efficient as possible. Right. Not everybody needs to understand what GIS is, but they do have a job
- [19:25]
A
to do and the configuration is reducing. Right. As we get further into the deployment of our AI strategy. It's really just a matter of creating a prompt that makes sense, you know, for the user's use case. So life gets simpler.
- [19:43]
B
We'll come back to AI. But you mentioned another thing. I think it's a critical thing to talk about when you're talking about, you know, any sort of 3D reconstruction or any, any Gaussian spotting is the images or the things that go into build that, right. And the example I used of, you know, the little demo I do, it's my dog on the couch, I'm doing a 360 and it's video. I ultimately chop it up into pictures and Walapu, it's done. But yours goes much beyond that, right? Like, tell me about the inputs, right? Are we talking lidar, are we talking like images, video? Like if you're used the example I think you said simic, right, to rebuild the, you know, the construction site. Like, what are they capturing? What is the frequency at which they're capturing?
- [20:27]
A
I'll start with like the source that comes from like a more traditional means, right? So we'll start with drones. So, you know, we have products like Reality Studio Drone to map. We have new capability, you know, that's like in Beta right now to create Gaussian splats from traditional drone capture data, right? So this is, you know, overhead, you know, aerial sources, not terrestrial sources right now. So it's being optimized for those tech stacks. You know, really, you know, the focus was to, you know, illuminate, you know, the detailed thin structures that we see in the environment, like particularly to help folks in the utilities industry, for example, to see like, you know, to find details of a substation, you know, that, that would be, that would be one example. And so, you know, we would take those images, you know, the still frame images, ingest those, do the processing to create the splat, and then we would bring that into the rest of the tech stack as just another example of our 3D tile format that can be pushed across our platform. On the more research side, the stuff that I'm more involved with, we're like leaning forward, looking at the terrestrial capture Data like from 360 cameras that you could run through a building, for example. And then, you know, using other projects right now to test like Stella V Slam to, you know, to pull poses and frames out of the video that we could then run into a pipeline for like radiance fields and Gaussian splatting. You know, we're really, really trying to converge the, the LIDAR and the, the Image based capture bits to create like high fidelity point clouds that we can create, you know, clean mesh data from for like the, the semantic representation of features, right? So taking really noisy input or intermediate output, like splats are beautiful, right? But you know, I need to do further processing to pluck out the bits that I care about, right? So if I, if I scanned a building and I want to know what the perimeter of the floor is and where the rooms are, that connotes a whole different line of processing. So, so that's where we start seeing this convergence between like the traditional lidar point cloud work, creating clean mesh data from that, segmenting it and then being able to pluck other things out of it from these beautiful high fidelity splats. Like now that I have the container, right, I want to get the stuff out of the container. Like where's the door, where's the window, where's the couch? Where's the dog on the couch? Right? And tell me what room the dog on the couch was in. And we want to try and make that as seamless and as automated as possible.
- [23:18]
B
I'm going to make an assumption based on what you said. Tell me if it's wrong. But when you're doing any sort of Gaussian spine, the image quality that you feed in and then the position of that to the subject is very important. What I think you're going to tell me is that, hey, we work with nerves, Gaussian splatter. But then where ESRI secret sauce is, is we're kind of regardless of what it's fed, it can make sense of it. Because I've done some Gaussian spine. That's freaking terrible. Like it looks like, you know, I'm like all gangly and stuff. But like, that's where it sounds like. This is where ESRI is proprietary in the sense where hey, feed it stuff, but kind of not whatever you put in. But you're going to have a highly likelihood of getting something out that's usable and workable, then running it through something else is. Am I hearing that correctly?
- [24:05]
A
That's like you're on track, right? Okay.
- [24:07]
B
See, hey, for a guy that knows nothing, I'm putting it together now.
- [24:11]
A
We wouldn't leave our users hanging that way, right? So there's no perfect way for splatting right now to not, not have, you know, shards and floaters, right? So, so if there are, you know, what do we do about that? We work with the sales team, we work with our deployment teams and our customer training about best capture techniques. We work with our partners and vendors who are doing the capture, you know, to provide that level of knowledge about like, you know, garbage in, garbage out, and then to provide tools to clean up artifacts that are going to be there regardless. Right. So there needs to be a clean pipeline. So what you're going to see in Reality Studio and the processing of this traditional drone capture data or the tool sets to do things just like that.
- [25:01]
C
I guess one of the important parts there too is that one thing that's unique that ESRI really brings to the table as well is that ability to position and locate and orient and scale that splat as well. Basically, building on it's not just capturing an asset in relative space, like we're talking about absolute space here. So how is that asset positioned? How is it oriented against other assets? Other sort of geospatial phenomenon as well. And so they think that's important too. And that's one thing that ESRI does, does best, basically.
- [25:30]
B
So it's the time and space of, you know, hey, there's the dog. Another bad example, the dog on the couch. But, you know, the couch is five feet long because the dog is three feet long or whatever it is. Right, okay.
- [25:41]
C
And ideally, where it is in space, right. It's, you know, that it's in your house. Your house is in wherever the city is on the surface of the earth within the universe.
- [25:48]
A
Well, what you're poking at is something that, that's a real concern for us and something that we're spending a lot of brain power on in terms of product development is, you know, how do, how do our customers leverage the legacy data sources that they've stewarded and maintained over time to contextualize, you know, the data that we're now creating more rapidly from via reality capture and AI technique. So that's, that's the magic, right? If our customers lose that investment, we are not doing our job.
- [26:21]
B
That makes total sense. Well, you, you said AI. I want to pull on that thread a little bit. It once again, going to go on a limb here is me, I get creating the 3D world, but where it's almost, I mean, and I hate to steal Nvidia's term here, but it's kind of the physical AI in the sense, but you're applying generative to it in the way where you've almost created a 3D world of, you know, my house on my block, in my city, in my state, you know, within the universe and all of that. But to be able to query that, whether that is, you know, Natural language to be able to understand is like, hey, what color is the brick? Or where are the two points of egress or ingress in the house? Is that also kind of what sets you apart? And I guess if that is the case, then how do customers kind of use that, that feature for generative to understand the models, like without them having to like, like where are some of the use cases for that?
- [27:18]
C
So I think some of the. I'll use one that's pretty, that's fairly simple that I know from a couple of years ago and that is around land use classification. So being able to take aerial imagery, if we start from that scale, aerial imagery, and instead of having a team of eight people sort of manually go through and try and figure out what portion within that image is a type of land use or is classified as a specific type of land use, like residential, commercial, industrial, different types of vegetation. It just depends on the level of fidelity that you're looking at there. I mean, that's one aspect of AI that's fairly simplistic that allows for, I think, you know, taking a team of, let's say eight people, you know, six to eight weeks to basically classify a scan, aerial scan, and sort of reduce that time to about maybe less than a day is something that's possible. Like you could do that. Right? And there's value there. It just means that you can work through larger amounts of data. Now granted, some of these things change over time too, so I think we have to factor that in as well. And so being able to process this over time for a more timely result is important. I think that's one aspect, if we think about it from a simplistic perspective. If you're looking at like AI, generative AI, from the perspective of, hey, I want to be able to get intelligence, I want to infuse more intelligence within my, you know, vector data, vector, vector based features. You know, we have the, I think it's looking at maybe just standards for how do we identify objects.
- [28:41]
B
Right.
- [28:41]
C
Object detection within these spaces. And so how do I know that there's a dog on the couch? Well, because I have, I can utilize AI in some capacity to look at different features of what I'm scanning and be able to identify that. Now is it perfect? No, but we, of course we train over time to be able to increase the validity of those scans in some aspect. But being able to capture that effectively, be able to store that in a way that the intelligence for those objects has then persisted in space and time and then accessible for you know, by our users. To be able to make decisions, I think is important in general.
- [29:12]
A
You know, part of the magic trick is, you know, I mean, this is probably the more deterministic part of it, rather than the AI algorithmic part is reprojecting, you know, what we find back into geographic space, right? So, you know, we have that covered, right? That's part of the magic sauce that ESRI produces. You know, another example that leads from there, like putting stuff in the right place in the world would be, you know, we have a capability, we call it oriented imagery. So this is non nadir imagery. Nadir would be like a flames, a flame, a plane is flying over, capturing images, right? And then non nadir would be like, you know, terrestrial. You know, I'm just holding my 360 camera or my iPhone and capturing stuff, right? So we can take like 360 video. Now we can use libraries like Stella Vslam, you know, leveraging AI and you know, some of these deterministic models to compute the camera pose positions from, from the video, right? And then we can reproject those positions back in a geographic space and then we can, we can take those 360 images and we can automatically populate these oriented imagery catalogs, you know, that have these kind of obliques, oblique views of the environment. And this is very powerful for our users. It's way faster than traditional means where I would had to hire a vendor to come on site, pay them half, you know, a quarter million bucks to run through with tripods and you know, like a ladybug camera system and capture all this stuff. Right now I have an untrained person running, zipping through a building with a GoPro 360. And now I can just run it through this processing to create beautiful 360 images that are located exactly, exactly where they need to be. I mean, we're talking about sub foot precision there. You know, on top of that we can create novel views, right? So if we're missing views, we can leverage, you know, the, the neural radiance fields to create, you know, position camera positions that, that weren't actually there in reality, right? So this is very, very powerful for our users for this, you know, kind of remote reconnaissance use case. Because you can't be everywhere all the time.
- [31:29]
B
Well, well, to get down to sub foot do you have to have a point of reference meaning, for example, you do a scan like you have to say, hey, this is, I mean, stupid example, but lat long or whatever you're doing, and then everything else there is mathematically probably calculated, but you have to have a point of reference on, correct?
- [31:47]
A
Yeah, yeah. So you do need ground control points, but that's already, you know, part and parcel of the ESRI stack, so. But it is true, you know, these newer capture technologies, they don't normally really care about, about XYZ and when they do, it's wildly inaccurate because they're in like urban canyon areas or in buildings and get a GPS signal there.
- [32:09]
B
Well, I know we've. We've been talking for a little over 35 minutes now, like, let's do this is we'll start with Rex and then we'll go to Pat and then we'll kind of wrap it up. Is, you know, from the. Obviously the space, you know, of reality capture three reconstruction has obviously evolved a lot over the last kind of few years. If you could share either, I guess you could pick one or both. Whatever you'd like is what do you see as kind of the next big thing that, you know, either maybe not market, but the next big use case that you'll see out of kind of ArcGIS or ESRI or where do you think the market might be in terms of a use case? Like a year from now?
- [32:47]
C
If we kind of build on the benefits of a game engine, right? Being able to bring in that, you know, simulation of the real world, but being able to bring in real world data and leverage it in that context, you know, bringing in atmospheres, physics, engines, special effects, animation, right? To bring your digital, basically your digital twin to life. Like your scene to life, your geospatial data to life. Like, where does that factor in? And so there's a couple things there. I'll say. The first one is being able to use a game engine, the benefits of game engine to provide that immersive application experience. That immersive application experience is probably going to be driven through either a handheld device or a headset. And we hear this a lot, right? You hear this a lot over the last. Gosh. I think we've been chasing this sort of virtual reality experience for decades, right? I mean, this goes back into like mid 20th century, but I think now we're at a stage where we're starting to see some advancements in hardware that can actually drive some of these experiences, these immersive experiences effectively. The first part there is like, how are folks using headsets with real world data? Right? Of course we see, you know, we have headsets that provide a fidelity that allows for that level of precision and interaction. Feels more natural. That'd be something like, let's say the Apple Vision Pro, right? Or you know, working with the varjo headset. Some great technology there, right? Some great advancements in those spaces. Now granted you're putting on a headset with purpose, right? This isn't a wearable actually. But if you're looking at working with high fidelity data and if you're working with, working in a geospatial context, you know, those devices are providing that level of interaction that really meets some of the requirements that have been established for decades that we're now seeing, you know, we're seeing come to life. And where are we seeing that? So one is in virtual reality. So it's inspecting remote locations with precision. It's also training workflows. So training workflows that allow for folks to be able to virtually go on site before they get to a site or being able to see what's happening before they're put into a situation. Now, because of the realism that's offered based on these experiences that are a lot of times driven by the fidelity that game engines can deliver, we see value, right? And so we're seeing a return on investment in that space. So we see advancements in those areas. Now kind of an aspect of that is what about the using a headset for mission planning, command and control, you know, situational awareness workflows. I know Pat mentioned earlier, working with law enforcement, we see a lot of engagements within defense, law enforcement, public safety around building a virtual tabletop experience so that they can collaborate with others that are remote. Not everyone has to be in the same room, but they can collaborate around a. Basically a digital twin on a virtual tabletop that is showing either real time data or maybe planning information going forward, maybe even, you know, simulation workflows. But they can collaborate around this virtual tabletop instead of having to, you know, extrapolate or at least abstract based on like some 2D map. They can actually virtually stand around each other in a. I think of it as like a virtual room around a virtual tabletop and they can see things happen real time. An example of that might be, let's say we're going to be fighting a fire. And so instead of actually physically crafting a sand table, which is what happens today, we already have all that digital information in a gis. Why don't we bring that to life in a way that we can have collaborators? Now granted, they're putting on a headset with purpose because they're planning how to attack or how to counter a fire. And so you have different individuals with different roles, pilots People on the ground, supervisors, they just think about this. They can gather around this virtual tabletop that they can change and move on the fly. They don't have to re sculpt and manually change this over time and place things. But if they can do that dynamically, they can see near real time information about where the fire's at, about where assets are on the ground, and then where they plan to be in the future. I think that allows for them to be more, to expedite those workflows in much to be able to handle those situations faster, make better decisions around what's happening on the ground. And so I think there's a, there's an aspect there that's just one example, but we see that within other situations to law enforcement and military as well. So just that that immersive experience there, as we begin to see that, you know, take root and come to life, we're seeing solutions going into, into practice now.
- [36:52]
B
Right.
- [36:52]
C
And we'll see those over the next year. If I build on that aspect. There is this and we talked a bit about precision of location, accuracy and orientation when you're in the field, whether it's capturing data or visualizing data. I think one thing that we're seeing come to life now is this basically positioning systems that allow for accurate location and orientation for off the shelf devices. And so I think something like a visual positioning system, obviously Google has one, Apple has one, Niantic has one. And being able to take those systems to be able to provide like basically centimeter level accuracy for consumer devices, I think that's an aspect there where it's not just for locating but also capturing data. And how do we do that in a way that's responsible, that allows for those devices to function in spaces where you may or may not have connectivity in some aspects. So that's a new area that we're trying to investigate to figure out how we can meet those needs. And there's a difference between public and private spaces, indoor and outdoor. Right. So that's a new area where we see some advancements. Then the last part there I just want to mention is on the AI front we talked a little bit about data capture and so being able to capture data from a variety of vantage points, whether on the ground, in the air. And I think one aspect there that's interesting about capturing meshes is that a lot of times you can't get basically a comprehensive view of assets. So if you're moving down a street and you're capturing, let's say, lidar Data or imagery data, you might not know what's behind the light post, what's behind the sign, what's behind a structure, but using AI, you might be able to assume what the back of a light post look like, what the back of a tree, or what the back of a sign or a bench or something like that looks like. And so looking at AI for data completion in some aspect, and so you don't have these gaps because you just weren't able to scan it from that specific side of that area. I think that's an interesting concept to be able to complete these scans in an intelligent way. And I think that's something that we're looking at, trying to figure out how we can enhance that so that we have a more complete data set or more complete data capture methods, methodologies going forward. So I'm kind of excited in that space too.
- [38:53]
B
Pat, do you have anything?
- [38:54]
A
Yeah, I just had a, had a few things. Some dovetail with what Rex was saying. So I think one of the things that I'm seeing is, you know, increased effort and need for this, you know, the rapid semantic contextualization of this reality capture content so that users can carry forward the investment they've already made in their legacy GIS data. So you're going to see more of that and more tools to do that. Lots of opportunities for partners and vendors in that space. Uh, number two, I mean, this just keeps becoming more, more and more the fact the world is the map now. So I'm seeing an increasing move in that direction. What do I mean? Like, rather than walking around life, you know, like this, staring at our phone, you know, we're going to see more, more and more applications and map development, like new types of cartography where, where we can curate the world as our map, right? Being being able to see what others can't in that sense. Right. So align with business needs, you know, through whatever wearable or even if I have to hold my phone up occasionally if I don't have a wearable to see those things. And, and that that's tightly connected, right, with this reality capture content and curating data sets on top of that, you know, and I think that there's like a couple different views of this. Like one pushes you more into holograms and VR and that's the kind of virtual tabletop from like the decision maker planner perspective and then the other, you know, and that supports like the virtual gatherings to make these decisions and then the other kind of pushes you back on the ground, right where I have my wearable and the world is the map. And so along all those lines, we'll just continue to see more progress, whether it's in the types of data that we're capturing to create these realistic representations that are more easily parcelable, you know, into like semantic chunks like, you know, building floor, window, campus, tree or whatever. Right. And then aligning that to our historical data and then fit for use with these different user types.
- [41:03]
B
So last question is, you know, someone's watching the episode. Where can they find you? Whether it be on social media, you know, corporate website, LinkedIn. We'll start with Pat, then we'll go to Rex and we'll wrap it up. Where can they find you? And learn more about ArcGIS and Esri.
- [41:17]
A
Yeah, I mean, my LinkedIn handle is pretty easy. It's just Patrick Wallace, so W A L L I S Wallace. And so look for my LinkedIn profile that way. You know, I don't, I don't have a separate practice page for, for our, our lab. So you just go to the esri.com site and dive in and take a look at our different product stacks and capabilities.
- [41:42]
B
All right. And to round us out, Rex.
- [41:45]
C
So LinkedIn as well, pretty easy to find out there. Feel free to reach out, connect and ask any questions you'd like there. Trying to be pretty active out there and publishing some of the work that our development teams are doing. Probably the best way to connect is through LinkedIn there. As far as discovering and working with our technology, most of our developer tools are free to get started. So if you go to developers archus.com that's a great place to get started. Our web native and game engine SDKs are available to get to start working with there at any time.
- [42:12]
B
So you heard here first. Developer.esri.com go get started. This is Logan. As I told you, it was going to be a little bit different today. Talking about, you know, from an isv, but learned a lot. You guys were great, great guests. Really appreciate you taking the time. And check out the show notes. You'll be able to get both pat's and Rex's LinkedIn account with a link to Ezra's website where you can get started. So with that being said, this is Logan from Reshaping workflows. Until the next time, keep all of your compute and AI running locally on Dell Pro Max and Nvidia RTX Pro and we'll see you on the next one.
- [43:04]
A
Do what you want. Do what you want.
- [43:06]
B
This podcast was produced in partnership with Amaze Media Labs.