Two college dropouts from the University of Pennsylvania have turned a ground-penetrating radar system into a startup that promises to help contractors avoid the multi-billion-dollar problem of accidentally severing underground pipes and power lines.
SubVysion, part of Y Combinator's most recent cohort, operates a small fleet of autonomous rovers that crawl across construction sites in Philadelphia and Texas, creating detailed 3D maps of what lies beneath. Contractors can then point their smartphones at the ground and see, through an augmented-reality overlay, exactly where a century-old water main or fiber-optic bundle sits in relation to a planned excavation.
The technology arrives at a precarious moment for the construction industry. Utility strikes reached a record high in 2025, with 221,717 documented incidents of workers accidentally hitting gas lines, water pipes, or electrical cables, according to the Common Ground Alliance, an industry safety group. The figure represents a steady climb from prior years, and the organization has estimated the broader societal cost at roughly $30 billion annually based on 2019 estimates.
Maxwell Stone and Wenson Tang, SubVysion's co-founders, spent their early twenties building software for other people's problems before deciding theirs was bigger. Stone had been writing machine-learning models at Roboflow, a computer-vision startup. There he deployed image-recognition systems across more than 10,000 pharmacies for a Fortune 100 healthcare company and worked on defect-detection algorithms for rail yards. Tang wrote emergency dispatch software at CentralSquare Technologies, a public-sector software firm. Both left school to join the accelerator.
Their pitch hinges on automation and precision. Traditional utility locators still rely heavily on operators pushing radar carts or electromagnetic wands across job sites, a slow process that often misses smaller lines or produces fuzzy depth readings. SubVysion's rovers, by contrast, work autonomously and layer ground-penetrating radar data with surface scans and visual positioning to generate what the company calls centimeter-accurate models. In an August launch post, Stone and Tang claimed their system delivers "100x better granularity and 10x the speed" compared to competitors, though they declined to specify which firms they benchmarked against or provide technical details that would let outside engineers verify the claim.
The company has published a handful of case studies that offer glimpses of the system in action. One project at Philadelphia's Pennovation Center, a university-linked research campus, reportedly uncovered multiple unknown water mains along a proposed pipe route. Another involved mapping utilities for a residential green-stormwater installation elsewhere in the city. Tang posted a video to LinkedIn in August showing one of the rovers, which the team nicknamed Suby, tracing utility lines at a Philadelphia site.
SubVysion enters a field that has begun attracting venture money and acquisition interest, though it remains fragmented between traditional survey-equipment vendors and newer AI-driven entrants. Exodigo, an Israeli-American firm, has raised undisclosed sums and in July acquired U.S. competitor V&A Consulting Engineers for a reported amount between $25 million and $30 million, according to Israeli tech publication CTech. The deal gave Exodigo a foothold in remote sensing and on-site reality capture for subsurface work. In the UK, RodRadar announced plans in July to install live-dig radar systems on more than 200 excavators operated by contractor Network Plus. 4M Analytics, which brands its offering as "America's Subsurface Model," has been posting updates about expanding nationwide coverage.
Established players like Leica Geosystems and IDS GeoRadar sell multi-channel ground-penetrating-radar arrays paired with survey-grade GPS positioning, equipment that can cost tens of thousands of dollars per unit. The American Society of Civil Engineers updated its subsurface utility engineering standard in 2022, codifying quality levels for investigation procedures, though the guidelines stop short of mandating specific accuracy thresholds.
SubVysion lists utility contractors, general contractors, civil engineers, and insurance companies among its target customers. The founders have indicated plans to move beyond individual job sites and pursue partnerships with municipalities and energy providers, a strategy that would require navigating procurement processes and regulatory approvals in dozens of jurisdictions.
For now, the company remains a two-person operation. Y Combinator's directory listed the team size at two as of late September, and public filings show a $350,000 convertible note dated March 26, according to CB Insights. SubVysion has not disclosed additional funding beyond its participation in the accelerator, and seed-round entries on platforms like Dealroom remain sparse on detail. The founders are betting that contractors weary of hitting pipes will pay for better maps, and that the cost of autonomous rovers will fall faster than the cost of repair crews and liability claims.
