A Chinese firm built a city traffic model on satellite imagery
A Shenzhen-listed location-data company decided the sensor city traffic was missing was never on the ground. Its answer went on show in Hangzhou this week.
Merit Interactive (每日互动), a Shenzhen-listed location-data firm, has released MUMO — Modeling of Urban Mobility — a city traffic model that fuses satellite remote sensing with ground-level traffic data. The company announced the model at the fifth Global Digital Trade Expo, which runs September 23–27 in Hangzhou and drew more than 2,000 exhibitors, roughly half of them AI-related. According to the company, high-resolution satellite imagery is spatiotemporally matched against multi-source ground data — intersection records and vehicle-feature data among them — so that isolated satellite snapshots become a continuous moving picture of a city's roads. The stated mechanism runs in four beats: continuous observation from orbit, real-time fusion on the ground, rolling model inference, and dynamic traffic-signal optimisation.
The product is a traffic-flow simulation model aimed at city traffic bureaus. Merit Interactive says a bureau can use it to rehearse "green wave" signal-coordination plans and quantitatively compare queue length against throughput across competing schemes, rather than trialling them on real roads. Chairman Fang Yi framed the pitch in the language of omniscience: with MUMO, he said, traffic police are like standing at a "god's-eye view," reading traffic from real data across both air and ground, which makes their dispatch decisions more precise. The company says it will keep exploring satellite data and space-based AI for traffic governance.
The reason this is interesting is that the bottleneck was never resolution. Commercial satellites have been good enough to watch a city for a decade; the problem is tempo — orbit over a given intersection every few hours, against traffic that changes every few seconds. A single overhead pass can tell you a junction exists and roughly how much asphalt it has. It cannot tell you how the queue behaves at 8:40am. MUMO's entire claim rests on the spatiotemporal matching step bridging that gap, and on enough ground data to interpolate honestly between passes. Contractors have chased this combination for years — aerial imagery as a cheap substitute for buried loop detectors and roadside cameras — because the sensor you don't have to dig up is a very attractive sensor.
Two things should temper the enthusiasm. First, the evidence is a launch-stage demo: Merit Interactive has published no accuracy figure, no simulation-versus-reality benchmark, and no named city running it in production. A "god's-eye view" is a vendor's metaphor, not a measured result, and a traffic model that is confidently wrong about queue length is worse than no model at all when a bureau uses it to retime a corridor. Second, satellite imagery degrades exactly when traffic modelling needs it most — cloud cover, night, and dense urban canyons all cut against the overhead view, which is presumably part of why ground fusion is doing so much of the work here. Treat the model as promising and unproven until a traffic bureau signs its name to a deployment. As we covered in August, China's AI ethics reviews reach city level — and a space AI model just trained itself in orbit, China's cities are the world's most eager buyers of AI built for municipal operations, which is precisely why the deployed numbers matter more than the launch copy. Merit Interactive is not a newcomer to this market: it has spent years selling location and mobility data to public and commercial customers, so MUMO is less a research bet than a packaging of data it already holds. The comparison worth keeping in mind is UrbanGround puts AI agents in a full-scale 3D Hong Kong — the other big bet on modelling a real city, built from the ground up instead of from orbit.
What to watch: whether MUMO gets a named city and a before-and-after number on signal timing. That is the only thing that separates a traffic model from a trade-show exhibit.
Would you trust a satellite-fed model to retime the signals on your commute? Tell us in the comments.
Sources: 证券时报·e公司 (Securities Times) · 第一财经 (Yicai)
