Seoul Robotics
Developing 3D LiDAR perception and infrastructure-based autonomy for industrial and logistics applications.
Website: https://seoulrobotics.org/
Cover Block
Open sources
| Field | Value |
|---|---|
| Name | Seoul Robotics |
| Tagline | Developing 3D LiDAR perception and infrastructure-based autonomy for industrial and logistics applications [Greenhouse, October 2026] |
| Headquarters | Seoul, South Korea |
| Founded | 2017 [Seoul Robotics, September 2022] |
| Stage | Series B [Seoul Robotics, September 2022] |
| Business Model | Hardware + Software |
| Industry | Deeptech |
| Technology | AI / Machine Learning |
| Geography | Global / Remote-First |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (3+) [Seoul Robotics, September 2022] |
| Funding Label | Series B |
| Total Disclosed Funding | Approximately $25,000,000 [Seoul Robotics, September 2022] |
Links
Open sources
- Website: https://seoulrobotics.org/
- LinkedIn: https://www.linkedin.com/company/seoul-robotics
- Careers: https://job-boards.greenhouse.io/seoulrobotics
What an Investor Needs First
PUBLIC Seoul Robotics develops 3D LiDAR perception software and infrastructure-based autonomous driving systems for controlled environments such as factories, ports, logistics yards, and airports, a positioning that merits investor attention because the company is arguing for autonomy at the site level rather than only inside the vehicle [Greenhouse, October 2026] [Greenhouse] [Seoul Robotics, September 2022]. The company traces its start to 2017, when it says four founders began working together in Seoul, and public materials now describe a broader business serving industrial and automotive logistics use cases rather than consumer autonomy [Seoul Robotics, September 2022] [LinkedIn].
The product case rests on two related architectures: a vehicle-centric system where autonomy runs on the vehicle, and an infrastructure-centric system marketed as Autonomy Through Infrastructure, where LiDAR sensors and site coordination software help direct vehicle movement across a facility [Greenhouse] [Greenhouse, October 2026]. That is a narrower but potentially more tractable commercial wedge than open-road autonomy, particularly in finished-vehicle logistics and yard operations, although most of the product detail in the current record remains company-described [Greenhouse] [LinkedIn].
On team quality, the public record is directionally positive but still thin. Hanbin Lee is listed as CEO and founder, with Crunchbase noting prior work as a self-driving car engineer mentor at Udacity, while LinkedIn-backed profiles identify Oran Kwon and Jaeil Park as founders and indicate Park's background in robotics and AI engineering [Crunchbase] [LinkedIn, retrieved 2026] [LinkedIn, retrieved 2026].
The financing picture is clearer than the operating metrics. Seoul Robotics announced a $25 million Series B in September 2022 led by KB Investment, with participation from Noh & Partners, FuturePlay, Korea Development Bank, Artesian, and Access Ventures, and the company describes itself as a hardware-plus-software business selling into enterprise and government-linked deployments [Seoul Robotics, September 2022] [Gaebler.com, September 2022] [ZoomInfo].
What bears watching over the next 12 to 18 months is whether the company can convert partnerships, customer logos, and hiring activity into independently visible deployments at scale. Public materials cite more than 80 customers as of September 2022, distribution partnerships in Japan, South America, Israel, and the United Kingdom, and 13 open roles in October 2026, but those signals are still weighted toward company-reported evidence rather than third-party operating disclosures [Seoul Robotics, September 2022] [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [Seoul Robotics, July 2021] [CDO Magazine] [Greenhouse, October 2026].
Claim stands unchecked -- This section relies materially on company publications and company-controlled recruiting materials, with partial corroboration for the Series B from Gaebler and founder identity from Crunchbase and LinkedIn.
Taxonomy Snapshot
| Axis | Value |
|---|---|
| Stage | Series B |
| Business Model | Hardware + Software |
| Industry / Vertical | Deeptech |
| Technology Type | AI / Machine Learning |
| Geography | Global / Remote-First |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (3+) |
| Funding | Series B, total disclosed approximately $25,000,000 |
Inside the Company
PUBLIC
Seoul Robotics presents as a South Korea based autonomy company built around a narrower, more industrial problem set than the consumer self-driving market usually implies. The company was founded in 2017 and is headquartered in Seoul, with public materials describing its focus as autonomous driving technology for business use cases rather than passenger ride-hailing or consumer vehicles [Crunchbase] [seoulrobotics.org].
The public record is clearest on the company’s positioning and later financing than on its earliest operating history. Crunchbase lists Hanbin Lee as CEO and founder, while company materials tied to the 2022 financing state that the business began in 2017 with four founders working in Seoul before expanding internationally over time [Crunchbase] [Seoul Robotics, September 2022]. The same company source frames the business around 3D LiDAR perception and infrastructure-based autonomy, which is a useful distinction because it places Seoul Robotics closer to industrial site automation than to general-purpose autonomous driving claims [Seoul Robotics, September 2022] [seoulrobotics.org].
The most clearly documented milestone in the available source set is the company’s September 2022 Series B. Seoul Robotics said it raised KRW 32 billion, approximately $25 million, in a round led by KB Investment, with participation from Noh & Partners, FuturePlay, Korea Development Bank, Artesian, and Access Ventures [Seoul Robotics, September 2022]. That round is also reflected by Gaebler, which reported a $25 million Series B on September 22, 2022, led by KB Investment [Gaebler.com, September 2022]. Subsequent public signals point more to continued operating build-out than to disclosed new financing, including a live Greenhouse board showing 13 open roles as of October 2026 across program management, machine learning, systems software, and automotive engineering functions [Greenhouse, October 2026].
Partially corroborated -- Core company facts are supported by Crunchbase and the company website, but several chronology details rely on company-published material.
Under the Hood
MIXED Product architecture and operating model
Seoul Robotics is positioning around a specific autonomy architecture rather than a general self-driving claim. Public materials describe the company as building 3D LiDAR perception and infrastructure-based autonomy for controlled industrial environments, including factories, ports, logistics yards, and airports [Greenhouse, October 2026]. The core product framing is "Autonomy Through Infrastructure," or ATI, which shifts sensing and coordination to site-installed infrastructure instead of placing the full autonomy stack on each vehicle; recruiting materials also describe both a vehicle-centric mode and an infrastructure-centric mode, suggesting the company can support operators that want either onboard autonomy or site-level orchestration [Greenhouse, October 2026].
The public record also points to a layered product set rather than a single application. ZoomInfo describes the company as supplying OEMs, system integrators, and government agencies through a proprietary perception platform called SENSR [ZoomInfo]. Separately, company and partner materials indicate geographic distribution for SENSR in Japan through Macnica, for SENSR and Discovery in South America through Dymaxion, and for SENSR-I in the United Kingdom through Digiflec [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [CDO Magazine]. Those references do not establish feature parity across products or the current revenue mix by module, but they do support the view that Seoul Robotics has commercialized perception software across multiple channels before and alongside its infrastructure autonomy push [GlobeNewswire, November 2022].
MIXED Deployment scope and technical signals
The strongest application evidence is in finished-vehicle logistics and yard operations. Public descriptions say the company automates finished vehicle logistics and truck yard operations, and its AVM, or Automated Vehicle Marshalling, platform is presented as enabling fleet-level autonomous movement inside automotive factories and large logistics hubs [LinkedIn] [seoulrobotics.tech]. In November 2022, GlobeNewswire reported that Seoul Robotics had commercialized ATI with a "Level 5 Control Tower," which the company describes as infrastructure-based autonomous driving for site operations [GlobeNewswire, November 2022]. A separate 2024 trade report says signals at a Salt Lake City installation were the first in the United States to be LiDAR-controlled, which is narrower than proving broad autonomous deployment but does show the company working at the intersection of roadside sensing and traffic control infrastructure [ITS International, July 2024].
The technical hiring pattern adds some limited color, with caveats. Current openings span machine learning, computer vision, system software, backend engineering, automotive systems, and vehicle electrical systems, which supports the basic picture of a company operating across perception, autonomy software, and systems integration rather than selling a standalone sensor tool [Greenhouse, October 2026]. Any deeper inference about the software stack remains just that, inferred from job postings, because the public materials here do not verify model architecture, simulation environment, remote operations layer, or safety-case design.
Claim stands unchecked -- This section relies heavily on company materials and recruiting pages, with selective support from GlobeNewswire, CDO Magazine, ITS International, and ZoomInfo.
Market Research
PUBLIC
The market matters now because Seoul Robotics is selling into a convergence of labor scarcity, industrial automation spending, and site-specific autonomy needs, but the public record here is notably thinner on third-party market sizing than on product intent [Seoul Robotics] [Greenhouse, October 2026].
The first constraint on market work is evidentiary. The supplied sources do not include a named third-party TAM, SAM, or SOM study for infrastructure-based autonomy, yard automation, or LiDAR perception software, so any hard market-size claim would overstate what the record supports. What can be said, conservatively, is that the company is positioned across several controlled-environment end markets that are large in operating spend even when autonomy budgets are not separately disclosed: factories, ports, logistics yards, airports, and finished-vehicle logistics operations are all named repeatedly in company recruiting and product materials [Greenhouse, October 2026] [LinkedIn].
That positioning matters because controlled sites are often earlier buyers of autonomy than open-road fleets. The company describes both a vehicle-centric system and an infrastructure-centric system, with its ATI model placing perception and coordination at the site level rather than solely on each vehicle [Greenhouse, October 2026]. In practical market terms, that suggests Seoul Robotics is not chasing the full consumer AV stack, but a narrower operational-automation budget inside industrial campuses, logistics compounds, and transport facilities where routes, safety procedures, and physical layouts are more bounded [Greenhouse, October 2026] [Seoul Robotics].
Public demand signals are present, but most remain company-reported. Seoul Robotics said in September 2022 that it was working with more than 80 customers in the 3D-perception industry, and its later hiring materials frame buyers around automakers, logistics operators, factories, ports, and airports [Seoul Robotics, September 2022] [Greenhouse, October 2026]. The macro logic is straightforward: where operators face driver shortages, repetitive low-speed vehicle movements, and pressure to improve throughput, an infrastructure-led autonomy system can be easier to justify than a full vehicle-by-vehicle retrofit, particularly if one installation can coordinate site traffic across multiple vehicles [Seoul Robotics] [Greenhouse, October 2026].
Adjacent and substitute markets broaden the opportunity set, but they also complicate category definition. Seoul Robotics appears to sit at the intersection of 3D LiDAR perception software, industrial autonomy, intelligent transportation infrastructure, and yard orchestration software, with distribution partnerships spanning Japan, South America, Israel, and the United Kingdom [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [Seoul Robotics, July 2021] [CDO Magazine]. That breadth is useful commercially because buyers may first engage through perception software such as SENSR before expanding into site-level autonomy, yet it also means the company competes for spend that could otherwise flow to vehicle OEM autonomy stacks, sensor vendors, warehouse automation systems, or traffic management infrastructure [ZoomInfo] [Greenhouse, October 2026].
Regulation and public-sector modernization are relevant tailwinds, though the evidence here is directional rather than exhaustive. Seoul Robotics says it serves OEMs, system integrators, and government agencies through its perception platform, and an ITS International report described a Salt Lake City installation whose signals were the first in the U.S. to be LiDAR-controlled [ZoomInfo] [ITS International, July 2024]. That does not establish broad regulatory adoption, but it does suggest an adjacent path into infrastructure upgrades where municipalities and transport agencies are already budgeting for safety, traffic control, and smart-roadway systems rather than for consumer AV deployment [ITS International, July 2024].
| Market slice | What the sources support | Evidence strength |
|---|---|---|
| Controlled-site autonomy | Factories, ports, logistics yards, and airports are explicit target environments | Company recruiting and website materials [Greenhouse, October 2026] [Seoul Robotics] |
| 3D perception software | SENSR is described as a proprietary perception software platform for LiDAR and 3D sensor solutions | Secondary company database plus distribution partnerships [ZoomInfo] [GlobeNewswire, April 2021] |
| Finished vehicle logistics | AVM is positioned around automating vehicle movement in factories and logistics hubs | Company product materials [seoulrobotics.tech] |
| Intelligent infrastructure | LiDAR-controlled signal deployment in Salt Lake City points to public-infrastructure adjacency | Trade publication coverage [ITS International, July 2024] |
The table shows a market with multiple budget owners rather than one clean software category. That can expand the reachable opportunity if sales execution is disciplined, but it also makes third-party sizing harder and raises the bar for showing repeatable deployment economics.
Claim stands unchecked -- This section relies primarily on company materials and recruiting pages, with limited third-party corroboration from trade and distribution announcements; no named third-party TAM/SAM/SOM report was provided in the source set.
Competition and Substitutes
MIXED Seoul Robotics appears to be positioning itself against a different part of the autonomy stack than most vehicle-centric autonomy vendors, with its public materials emphasizing infrastructure-based control for constrained industrial sites rather than onboard autonomy for open-road driving [Greenhouse, October 2026] [Greenhouse] [Seoul Robotics, September 2022].
The competitive map is therefore best read by segment, not by a single peer set. In finished-vehicle logistics, truck yards, ports, and factory compounds, the practical alternatives include incumbent manual operations, site-specific automation programs run by OEMs or systems integrators, and vehicle-centric autonomy stacks that place sensing and compute on each machine or car [Greenhouse, October 2026]. Seoul Robotics' own framing draws this distinction directly, describing both a vehicle-centric system and an infrastructure-centric system, with its ATI approach built around LiDAR perception, site-level coordination, and autonomous vehicles [Greenhouse]. That matters because the company is not merely competing with another startup's software, it is also competing with a buyer's decision to automate the site through labor, process redesign, conventional yard software, or OEM-led internal programs.
The company's edge, on the evidence available, is clearest in system architecture and channel breadth, though both advantages should be treated cautiously because much of the underlying proof is company-reported. Seoul Robotics says it serves factories, ports, logistics yards, and airports, and public materials show a pattern of distributor and regional partnerships spanning Japan through Macnica, South America through Dymaxion, Israel through Hyper-Tech Advanced Systems, and the United Kingdom through Digiflec [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [Seoul Robotics, July 2021] [CDO Magazine]. If that channel structure is active rather than merely announced, it suggests a go-to-market model that can travel through local industrial and sensor ecosystems rather than requiring the company to build every market directly. The durability question is whether this edge compounds into deployment data, integration know-how, and reference sites, or whether it remains perishable because distributors can switch suppliers and infrastructure-heavy projects often have long sales cycles and bespoke implementation work.
The exposure is equally specific. Seoul Robotics does not appear, from the cited public record, to own a proprietary vehicle platform, a captive OEM channel, or a public financing base comparable to the better-capitalized autonomy categories that can subsidize long enterprise pilots [Seoul Robotics, September 2022] [Gaebler.com, September 2022]. Its disclosed Series B was $25 million in September 2022, led by KB Investment, which is meaningful but still modest for an autonomy company spanning hardware, software, and industrial deployments [Seoul Robotics, September 2022] [Gaebler.com, September 2022]. In that sense, the company's strongest wedge is also its most obvious objection surface.
The most plausible 18-month scenario is a bifurcation between sites that want centralized, infrastructure-led orchestration and sites that prefer vehicle-embedded autonomy. Seoul Robotics is the likely winner if industrial buyers, especially automakers and logistics operators, prioritize fleet-level coordination in controlled environments and accept fixed-site sensor installation as the cleaner operating model [Greenhouse, October 2026] [seoulrobotics.tech]. Seoul Robotics is the likely loser if those same buyers decide the retrofit cost, procurement complexity, or systems-integration burden of infrastructure autonomy outweighs the benefits, in which case incumbent manual workflows or OEM-managed vehicle-centric programs keep the upper hand. With no named competitors verified in the source set, the public evidence supports a narrower conclusion: the contest is presently between architectural approaches and distribution execution, not between clearly documented startup peers.
Partially corroborated -- Relies on company materials and one trade source for product architecture and partnership footprint, with partial corroboration of funding by Gaebler.com;
Opportunity
Upside Case
PUBLIC The prize here is unusually large if infrastructure-based autonomy proves easier to deploy and safer to operate than putting the full stack on every vehicle, because that would position Seoul Robotics as a control layer for high-throughput industrial sites rather than as a single-feature autonomy vendor [Greenhouse] [GlobeNewswire, November 2022].
The headline opportunity is straightforward: Seoul Robotics could become a default infrastructure software and perception provider for controlled-environment vehicle movement in factories, logistics yards, ports, and airports, where routes are repetitive, operating domains are bounded, and labor shortages create a clear economic reason to automate [Greenhouse] [seoulrobotics.org]. That outcome is still conditional, but it is not purely aspirational on the evidence available. The company has publicly described a product architecture that separates infrastructure-centric autonomy from vehicle-centric autonomy, has commercialized its Autonomy Through Infrastructure system with what it calls Level 5 Control Tower, and has built distributor relationships across Japan, South America, Israel, and the United Kingdom, which suggests an effort to scale through channel access rather than direct deployment alone [Greenhouse] [GlobeNewswire, November 2022] [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [Seoul Robotics, July 2021] [CDO Magazine]. Seoul Robotics also reported more than 80 customers in the 3D-perception industry as of September 2022, although that figure is company-reported and does not establish production-scale autonomy revenue by itself [Seoul Robotics, September 2022].
The most plausible paths to scale are not all the same, which matters because this category is still being defined. One path is to become the standard operating layer for finished-vehicle logistics and yard automation, where the company already frames its Automated Vehicle Marshalling product and truck-yard automation work [LinkedIn] [seoulrobotics.tech]. A second path is to become the international LiDAR perception and orchestration stack distributed through regional partners, using SENSR and related products as the entry wedge [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [CDO Magazine]. A third is to use a high-visibility infrastructure deployment, such as the Salt Lake City lidar-controlled traffic signal installation cited by ITS International, to strengthen the case that site-side intelligence can control more than one vehicle class or operating environment over time [ITS International, July 2024].
| Scenario | What happens | Catalyst | Why it's plausible |
|---|---|---|---|
| Industrial site standard | Seoul Robotics becomes a repeat vendor for factories and logistics yards that need autonomous vehicle movement inside controlled sites | More ATI or AVM deployments with automotive and industrial operators | The company already positions ATI and AVM for factories and logistics hubs, and recruiting materials cite relationships with major automotive and industrial names [Greenhouse] [seoulrobotics.tech] |
| Channel-led global expansion | Regional distributors turn SENSR and related software into a lighter-capital route for international market access | Expanded distributor execution in Japan, South America, Israel, and the UK | Publicly disclosed distribution partnerships show the company has already tested this go-to-market motion across several geographies [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [Seoul Robotics, July 2021] [CDO Magazine] |
| Infrastructure autonomy proof point compounds | A few visible infrastructure-controlled deployments persuade operators that site-level autonomy can outperform vehicle-by-vehicle retrofits in bounded domains | Reference deployments such as the Level 5 Control Tower commercialization and lidar-controlled signal work in the US | Seoul Robotics has publicly tied its differentiation to infrastructure control, and third-party coverage points to at least one US installation using lidar-controlled signals [GlobeNewswire, November 2022] [ITS International, July 2024] |
The compounding logic rests on reuse. If an operator installs site-side perception, coordination software, and operating procedures once, the next expansion can attach additional routes, vehicle types, or adjacent yards without rebuilding the entire autonomy stack from scratch, at least in theory [Greenhouse] [GlobeNewswire, November 2022]. That creates a plausible flywheel: one deployment generates site data and operational know-how, which improves perception and orchestration performance, which makes the next deployment easier to sell, especially if distributors and system integrators can package the software into repeatable local offerings [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [CDO Magazine]. The evidence that this flywheel has started is still modest and mostly company-led, but the combination of more than 80 reported customers in 2022, a stated focus on OEMs, integrators, and government agencies, and continued hiring across machine learning, autonomy software, systems engineering, and program management is at least consistent with a company building for repeat deployments rather than one-off pilots [Seoul Robotics, September 2022] [ZoomInfo] [Greenhouse, October 2026].
The size of the win cannot be pinned to a hard market value from the cited record because the source set does not include a verified public-peer market cap, acquisition multiple, or named TAM study. Even so, the upside frame is still clear enough to state conservatively: if Seoul Robotics were to become a category-level infrastructure autonomy platform for industrial vehicle movement across multiple geographies, that would support venture-scale outcomes rather than niche software economics (scenario, not a forecast) [GlobeNewswire, April 2021] [GlobeNewswire, May 2021] [GlobeNewswire, November 2022]. The constraint is not whether the operating problem exists. The public record already shows the company aiming at persistent pain points such as driver shortages and low-speed, repetitive vehicle movement in controlled environments. The real question is whether infrastructure-based autonomy becomes the preferred deployment model before larger autonomy, industrial automation, or OEM-led systems absorb the same budget line [seoulrobotics.org] [Greenhouse].
Claim stands unchecked -- This section relies heavily on company materials and recruiting pages, with limited third-party corroboration from GlobeNewswire, CDO Magazine, and ITS International.
Sources
Open sources
[Greenhouse, October 2026] Seoul Robotics job board | https://job-boards.greenhouse.io/seoulrobotics
[Seoul Robotics, September 2022] Celebrating Seoul Robotics’ Series B Funding with a word from our CEO, HanBin Lee | https://seoulrobotics.org/blog/blog-post-title-two-mdn8d-mrkl9-5jfwc-dznjx-5gz8l
[Greenhouse] Project Manager - Autonomous Driving Programs | https://job-boards.greenhouse.io/seoulrobotics/jobs/5385714008
[LinkedIn] Seoul Robotics | LinkedIn | https://www.linkedin.com/company/seoul-robotics
[Crunchbase] Seoul Robotics - Crunchbase Company Profile & Funding | https://www.crunchbase.com/organization/seoul-robotics
[LinkedIn, retrieved 2026] Oran Kwon님 - Seoul Robotics | https://kr.linkedin.com/in/orankwon
[LinkedIn, retrieved 2026] Jaeil Park - Netropy AI | LinkedIn | https://www.linkedin.com/in/jaeil-park-6468a6a8/
[Gaebler.com, September 2022] Seoul Robotics Closes $25 Million Series B Financing | https://www.gaebler.com/News/Finance/Seoul-Robotics-Closes-25-Million-Series-B-Financing-n22317
[ZoomInfo] Seoul Robotics - Overview, News & Similar companies | ZoomInfo.com | https://www.zoominfo.com/c/seoul-robotics/540754782
[GlobeNewswire, April 2021] Seoul Robotics Announces New Partnership With Macnica to Bring Cutting-edge 3D Perception Software to the Japanese Market | https://www.globenewswire.com/news-release/2021/04/20/2213054/0/en/Seoul-Robotics-Announces-New-Partnership-With-Macnica-to-Bring-Cutting-edge-3D-Perception-Software-to-the-Japanese-Market.html
[GlobeNewswire, May 2021] Seoul Robotics Signs New Distributor Dymaxion for South American Expansion | https://www.globenewswire.com/news-release/2021/05/20/2233297/0/en/Seoul-Robotics-Signs-New-Distributor-Dymaxion-for-South-American-Expansion.html
[CDO Magazine] Seoul Robotics Enters UK Market Through Digiflec Partnership | https://www.cdomagazine.tech/2021/08/04/seoul-robotics-enters-uk-market-through-digiflec-partnership/
[GlobeNewswire, November 2022] Seoul Robotics Commercializes Level 5 Control Tower for Infrastructure-based Autonomous Driving | https://www.globenewswire.com/news-release/2022/11/16/2557014/0/en/Seoul-Robotics-Commercializes-Level-5-Control-Tower-for-Infrastructure-based-Autonomous-Driving.html
[ITS International, July 2024] LiDAR-controlled signals make U.S. debut in Salt Lake City | https://www.itsinternational.com/its1/news/lidar-controlled-signals-make-us-debut-salt-lake-city
[seoulrobotics.org] Seoul Robotics : B2B Autonomous Driving Technology | https://seoulrobotics.org/
[seoulrobotics.tech] Seoul Robotics : B2B Autonomous Driving Technology | https://seoulrobotics.tech/
[Seoul Robotics, July 2021] Seoul Robotics Announces New Distributor in Israel | https://seoulrobotics.org/blog/seoul-robotics-announces-new-distributor-in-israel
Articles about Seoul Robotics
- Seoul Robotics Puts Autonomy in the Infrastructure, Not the Vehicle — The Korean startup’s approach flips the self-driving script by installing the brains of the operation in the site itself, a bet that has landed it with BMW and Mercedes.