[ENG][DAN 24] KEYNOTE 02: Spatial AI
Hello! I’m Saehoon Lee, and I oversee business planning for NAVER Place and Map. To give you some context, I work on NAVER’s space-related services, which brings us to today's topic: the space we’re in right now. Let me tell you how spatial AI - AI embedded in spaces - will transform the integration of online and offline experiences in the future. This session on spatial intelligence is divided into two parts. First, I'll explore how NAVER Map merges with spatial intelligence to reshape our daily lives.
Then, we’ll dive into the technologies driving these changes and the experiences and future they aim to build. As we begin with NAVER Map and its impact on everyday life, let’s take a step back into the space you were in this morning. That was a quick video of the journey from COEX to the Grand Ballroom.
The GBC construction is full swing and new shops have opened on every street. Even this venue hosted a different event just yesterday. It’s clear that offline spaces are changing faster than we might imagine.
NAVER Map plays a crucial role in bringing these fast changes in offline spaces to the digital world. I’m sure many of you have experienced the frustration when online information doesn’t match reality. That’s why NAVER Map is committed to its core value of providing information that stays true to reality as much as possible, and I believe what enables this is spatial intelligence. Since the early days to the present, NAVER Map has been committed to capturing accurate offline information.
NAVER Map started with now-essential services like Street View and public transportation information, later upgraded to include navigation and saving features. Today, it offers business-specific details and even facilitates transactions such as real-time reservations and bookings. With these updates, NAVER Map has become an indispensable part of our daily lives, used by 26 million users every month. The future of NAVER Map, redefined by spatial intelligence, boils down to three key words: Real, Now, and For You. Spatial information will be so lifelike that it feels real, even if you’re not there in person.
Time information will be delivered in real time, making it available when users need it most. As spatial intelligence integrates with personalized recommendations, it will be easier to discover desired locations such as restaurants and cafes on the map. I’m sure many of you attending DAN 24 today used NAVER Map to get here. Let’s retrace the journey from home to COEX as I introduce the upcoming changes to NAVER Map. The scenario has three steps: preparing before leaving home, traveling to the venue, and experiencing the location upon arrival.
Let me walk you through each step. First, before heading out, you check out the COEX area using the 3D Street View. What makes it special is that, unlike the previous version, it doesn’t just guide you.
It also overlays information onto the three-dimensional space. For example, you can see nearby locations and even menu options or discount coupons for cafes, adding a sense of realism. There’s the DAN 24 venue, our destination. Let's click on it. It provides a wealth of information about DAN 24. At the bottom, you can explore different spaces - through the VR Indoor Tour. This VR Indoor Tour technology offers an immersive experience in 3D that feels like being right there in person, without a physical visit.
Here’s footage from the recent Aladdin musical pop-up store. If you haven’t visited a pop-up, the VR Indoor Tour lets you preview the space and its exhibits. For those who've already been there, it’s a great way to revive the memory and experience. Although the Aladdin pop-up store has closed, the fact that it remains as spatial information we can revisit signifies the value of spatial intelligence. Now, let’s head out and make our way to COEX. Along the way, you’ll get to experience features like navigation, public transportation information, indoor maps, and AR navigation.
Just as you step outside, a notification pops up. Naver Map lets you know when to leave if you’re driving, based on the schedule information you registered. If you go ahead with the navigation, it also checks whether you can arrive before the event begins and reminds you of the event’s start time while driving. As you head out, you set COEX as destination and start receiving directions for public transportation. The app notifies you where to transfer. Even if you’re busy with music, news, or something else,
it alerts you to get off as you approach Bongeunsa Station, and even specifies the exit number. And here we are, in front of Exit 7 at Bongeunsa Station. Next up is the Indoor Map feature, which guides you from Bongeunsa Station to the COEX Grand Ballroom. In a large, multi-level complex like COEX, with four underground and five above-ground floors, it can be confusing to decide whether to head to the surface or use underground passageways.
The Indoor Map eliminates this uncertainty by providing clear directions. While some may be familiar with COEX, for others, finding their way to the destination can be a real challenge. Now, let’s finish the journey using AR navigation. Inside a labyrinth-like environment, it works like a car navigation, leading you directly to the destination. It gives you clear instructions as you weave through shops on the basement floor and even tells you where to turn. After climbing the stairs to the first floor, you’ve safely arrived at today’s venue, the Grand Ballroom.
Now that you’re at the venue, you'll naturally check online information about the DAN 24 pop-store and explore place recommendations to discover popular spots nearby after the event. With NAVER Map, we’ve safely made it to the DAN 24 event venue. Next, let me introduce the pop-up store service, which digitally showcases today’s event on NAVER Map. Using spatial intelligence technology, this service archives short-lived pop-up stores online. Let me open the DAN 24 page and check out the live reactions coming in from the venue.
Feel free to take out your phone, open the NAVER Map app, and take a look for yourself. To finish our scenario, let’s look for a restaurant near Samsung station after the event. In the updated NAVER Map app in the future, popular nearby places will be marked on the map, along with real-time insights about which areas are trending and where it’s unusually crowded. Also, the Nearby Popular Spots section shows consistently trending locations to help you pick a restaurant.
Further down, a feed-style area provides personalized recommendations based on your saved and clicked interests, featuring review videos for a more vivid and engaging experience. So, how was the journey of connecting offline and offline worlds with NAVER Map? As mentioned in the beginning of the presentation, spatial intelligence envisioned by NAVER Map focuses on capturing everything happening in a space realistically and in real time. And this is not limited to NAVER Map. It is being applied across many other NAVER services as well. Let me share another example of spatial intelligence in action.
For those who’ve taken a day off to check out real estate listings, NAVER Real Estate offers a better solution. Powered by spatial intelligence, it provides the virtual layout and spacing of apartment buildings. With 3D models of interiors and detailed floor plans, you can immerse yourself in the space and make informed decisions, all without the need for an on-site visit. Spatial intelligence can expand beyond Map and Real Estate to various NAVER services. It can be applied not just to residential spaces but to the entire physical world we interact with.
The integration of Spatial AI and NAVER will empower us to experience so much more without ever psychically being there, seamlessly embedding itself into our daily lives. And because it's NAVER, the imagination doesn’t stop here. From here, Donghwan Lee, Leader of the Vision Group at NAVER Labs will take over to explain the spatial intelligence technologies behind our services and their future potential.
Hello, I'm Lee Donghwan of NAVER Labs. How did you feel about NAVER’s new services we presented earlier? Now, let’s dive into NAVER’s spatial intelligence technologies that enabled these innovations. So, what exactly is spatial intelligence? Space is a world made up of infinite data, and spatial intelligence is the tool to explore and connect this space.
It’s not so different from the online search and services NAVER has previously provided. Except, it extends into the realm of physical spaces. Spatial intelligence is rapidly emerging as a key focus in the global tech world, following language models.
Recently, the global market research firm Gartner projected the spatial computing market will grow more than tenfold, reaching $1.7 trillion by 2033. NAVER has been proactively investing in core spatial intelligence technologies for years. We have undertaken numerous research projects and also achieved world-class capabilities across all areas. With a unique strategy, we’ve built a robust ecosystem for spatial intelligence technology. Our process begins with the creation of the digital twin that precisely mirrors the real world. Then, we connect reality and the digital world using AI and cloud systems, ensuring real-time synchronization between the two.
This technological foundation drives the development of robots, autonomous driving, and AR services that can interact with the physical world. Digital twins, in particular, builds the stage where spatial intelligence comes to life. At NAVER, we have developed our own mapping devices to create digital twins of nearly all everyday spaces. The M Series includes mapping robots, the T Series features wearable devices such as backpacks, and the R and P Series can be mounted on vehicles. We also rely on aerial photography. While our team doesn’t build airplanes, we develop AI capable of processing thousands of aerial photos to create 3D models and realistic virtual environments.
The video you’re watching is an example of a digital twin. At first glance, it might look like drone footage, but if you look closer, you’ll notice it’s a 3D-rendered virtual world. Creating digital twin data for smart buildings is also straightforward. What you’re seeing here is a 3D digital twin model of NAVER’s second headquarters, 1784. The R Series collects 3D road data and is used to create HD maps, which are crucial for autonomous vehicles.
Now, you’re looking at the P Series. P stands for panorama. This series is specialized in gathering 3D panoramic data. This data is also utilized in NAVER Map's 3D Street View service, which we introduced earlier.
We’ve been developing our own devices and AI capable of creating digital twins for all types of everyday spaces, from buildings to roads and even entire cities. This full-stack digital twin solution is a rarity even on the global stage. Beyond digital twins, NAVER also excels in another key area: positioning, or location recognition. Here, NAVER has focused extensively on visual localization, an AI technology that can pinpoint exact locations with just a single photo. Here’s a video introducing this fascinating technology. Our AI-enabled positioning technology is already commercially available on the NAVER Cloud Platform as a product called ARC Eye.
This video shows how ARC eye creates AR navigation for a theme park. It can scan an entire theme park in just two and a half hours and develop an AR navigation app in three days, delivering incredible accuracy and efficiency. This AR navigation provides accurate directions even in the most complex indoor and outdoor spaces. The possibilities extend far beyond navigation. Imagine viewing ancient artifacts in a more immersive way in a museum. Or you can simulate interior designs for buildings that are still under construction.
All the AR videos you've seen so far are actual footage captured on-site. Robots also represent a major opportunity for spatial intelligence. Just as GPS made car navigation an essential part of everyday life, spatial intelligence will do the same for robot services. Visual localization connects digital twins with the real world, equipping robots with spatial intelligence. Since it only requires a camera, it significantly lowers production costs and reduces the technical barriers to autonomous robot navigation.
In addition, companies that produce robot services can focus on their services because Naver's solutions can be used in areas that require a lot of technology, such as mapping and positioning technology When talking about spatial intelligence, there’s an important point I’d like to highlight. This graph shows the cumulative number of NAVER’s spatial intelligence patents, which has hit 521 this year. They make up over 60% of NAVER Labs’ research-related patents. You may notice a spike around 2021.
What happened at NAVER back then? The completion of our testbed. A testbed is a crucial competitive edge when it comes to spatial intelligence research. In 2021, NAVER built the world’s first robot-friendly building. This 28-story structure with 8 underground levels has become a hub for our spatial intelligence research activities. With countless people, diverse infrastructure, and over 100 robots, our technology evolves rapidly in real-world environments.
And last year, we added another testbed spanning the size of 41 soccer fields. While 1784 serves as a vertical testbed, the data center GAK Sejong is a horizontal one, acting as a base for developing smart city technology including autonomous driving. The video shows robot perception research designed to help robots recognize people and understand their movements. An essential technology for robots to coexist naturally with humans. We are also creating virtual simulation environments for robot services at 1784. We're digitizing the surrounding areas into a digital twin and exploring ways to reconstruct videos from any angle.
In this virtual space, camera movements are not limited by physical constraints. Enabling the creation of entirely new videos from any viewpoint, this technology has vast potential across industries such as film, gaming, and media. Imagine how convenient this could be for navigating to a destination. Our AI research focuses on foundation models. Foundation models are pre-trained on massive datasets and form the backbone of various technologies like GPT, a popular language generation model.
NAVER has been leading the way, transitioning all spatial intelligence research to foundation models as early as 2021. 'Dust3r', the groundbreaking AI we introduced this year, is an example of our foundation model-driven advancements. This AI turns photos into a 3D digital twin in just a few seconds. Even with a single photo. While technologies I described earlier require experts with specialized devices and knowledge, Dust3r democratizes spatial intelligence.
For example, with Dust3r, you can take photos of your room and effortlessly turn it into a VR space. You can even preserve 3D memories of the place you once lived in. Once confined to 2D, our memories can now become more vivid and nostalgic in 3D. Many more spatial intelligence technologies are being developed and enhanced at NAVER’s testbed. Naver is gradually incorporating these technologies into its services.
Examples include the 3D Street View service, virtual VR tours, real estate complex tours and property tours that were mentioned earlier. Our vision is much bigger. Companies and organizations around the world can become partners in our spatial intelligence ecosystem. Here’s a short video to introduce this idea. The NAVER TwinXR platform brings together NAVER’s digital twin solutions and a wide range of spatial intelligence technologies. From immersive experiences via smartphone or AR glasses to robotics, autonomous driving, and smart city infrastructure, the TwinXR platform makes spatial intelligence accessible to everyone.
The platform is already collaborating with leading global partners. Among them is Swiss-Mile. This world-class robotics company is working with us to apply NAVER’s digital twin and positioning systems to quadruped robots. This exciting project explores how robots can deliver services in dynamic environments like construction sites. Another significant partner is NTT Group, Japan's top telecom company.
With NTT East, which covers Tokyo and eastern Japan, we plan to collaborate in areas like smart building, robotics, and AR. You’ve probably seen this in the news, but this is the first time we're revealing actual visuals. As you can see, the Saudi Arabia Digital Twin Platform project is coming along smoothly. The footage shows a flood simulation with a virtual model of Jeddah, Next is a 3D city model of Medina, where simulations for urban planning can be conducted. You can even take a virtual tour of the luxurious residential complexes developed by the Saudi Ministry of Housing. Our spatial intelligence is setting a new global standard for future cities.
What you’ve seen today is the tip of the iceberg. We look forward to sharing more exciting news with you more frequently. NAVER is a clear leader in spatial intelligence, offering both “everyday spatial services” and “spatial technologies of tomorrow.” Our mission is to transform spaces from stages for life into active platforms where information and services blend seamlessly to create new experiences. Information will reach you precisely when and where you need it, online and offline services will integrate effortlessly, and immersive experiences will be taken to new heights. With this vision, we will continue to innovate and take on global challenges.
We look forward to your continued support. Thank you.
2024-12-21 00:30