--- title: "China's \"vehicle-road-cloud\" bet: 20 cities rewire the street for self-driving" date: 2026-10-22 category: Policy & Governance site: NeuroAI canonical: https://neuroai.site/a/na-policy-vehicle-road-cloud-pilot language: en --- # China's "vehicle-road-cloud" bet: 20 cities rewire the street for self-driving > A 2024–2026 pilot has 20 Chinese cities build roadside sensors and cloud systems so autonomous vehicles can safely share the street. A driverless shuttle waits at a city intersection. Its own sensors can see the crossing, but they cannot see the truck hiding two blocks away behind a curve — a blind spot no onboard camera will ever fix. That gap is the whole problem China's latest connected-vehicle pilot is trying to close. Instead of asking every vehicle to become a perfect, all-seeing robot, the program asks the city itself to help. The street becomes a participant. Poles carry sensors; a city-level cloud fuses what they see; the car receives a warning it could never have computed alone. ## What the pilot actually is In early 2024, five central ministries opened an application pilot for what China calls "vehicle-road-cloud integration" (车路云一体化). The name is a compact description of the architecture: the **vehicle** (the car), the **road** (roadside infrastructure), and the **cloud** (a city-level control platform) are designed to work as one system rather than as separate products. The pilot runs for roughly three years, through 2026. It is not a subsidy program and it is not a ban. It is a coordinated field test at city scale, meant to turn scattered demonstration zones into something connected and repeatable. ## Who signed it and when The notice was issued by the Ministry of Industry and Information Technology together with the Ministry of Public Security, the Ministry of Natural Resources, the Ministry of Housing and Urban-Rural Development, and the Ministry of Transport. The document inviting cities to apply appeared in mid-January 2024, and the first batch of selected cities was announced on July 3, 2024. Those five agencies matter because the system cuts across their mandates: transport builds the roads, public security manages the vehicles and safety, natural resources handles mapping, housing and urban-rural development handles street infrastructure, and industry and information technology coordinates the technology stack. ## The nine jobs The pilot assigns nine concrete work items. The headline ones: - Build **intelligent roadside infrastructure**, with full 5G coverage in pilot areas and cellular vehicle-to-everything (C-V2X) equipment deployed along the routes. - Retrofit traffic signals and signage so they are networked, targeting a networking rate above 90 percent. - Raise onboard terminal fitting rates, with every pilot-run vehicle required to carry a C-V2X terminal and a digital identity certificate. - Stand up a **city-level service and management platform** that can absorb roadside and vehicle data. - Run scaled demonstrations across buses, passenger cars, automated parking, urban logistics, and delivery. - Explore safe use of high-precision maps. - Develop the standards, testing, and evaluation system. - Build a cross-domain identity recognition framework. - Improve road-traffic safety and probe new investment, construction, and operation business models. ## The twenty cities The July 2024 list named twenty cities or city groups: Beijing, Shanghai, Chongqing, Ordos, Shenyang, Changchun, Nanjing, Suzhou, Wuxi, Hangzhou–Tongxiang–Deqing, Hefei, Fuzhou, Jinan, Wuhan, Shiyan, Changsha, Guangzhou, Shenzhen, Haikou–Sanya–Qionghai, and Chengdu. The mix is deliberate. It spans China's richest megacities and smaller industrial bases, coastal and inland, so the lessons are not tied to one kind of road network or one climate. ## What builders must install The technical floor is specific. Public-sector fleets — buses, official cars, taxis — are encouraged to retrofit C-V2X terminals, and newly sold vehicles in pilot areas are expected to reach a 50 percent terminal-fitting rate. Every vehicle that actually runs in the pilot must carry the terminal and a digital identity. On the roadside, the expectation is that key intersections get sensing devices and edge compute (算力) nodes that talk to the city platform. The point is to shrink the blind spots a car cannot see for itself and to give the cloud a chance to warn the vehicle in time. ## The money behind it This is not cheap. Beijing's own pilot infrastructure tender, opened in late May 2024, carried a total investment of about RMB 9.94 billion (≈ US$1.4bn / HK$10.9bn), covering thousands of intersections across more than a dozen districts. Wuhan and other cities followed with projects of their own. The sums signal that "integration" is treated as civil infrastructure — closer to transit than to a consumer gadget. ## Why it matters beyond China Most self-driving narratives are car-centric: make the vehicle smarter. This pilot bets the opposite is also necessary — make the environment legible to machines. For global operators, the takeaway is practical. A company bringing autonomous or connected vehicles to China will eventually be expected to plug into local roadside and cloud systems, not just ship a clever car. It also sets a template other countries are watching: whether city-scale, infrastructure-assisted autonomy is safer or merely more complex than the vehicle-only approach remains an open engineering question. ## Honest limitations This article describes a policy framework and its stated goals, not a measurement of results. As of this writing, the pilot is ongoing and no single official scorecard confirms every 2024–2026 target has been met. The specific fitting-rate and networking-rate figures are the program's own requirements, not independently verified outcomes. The Beijing investment number comes from a public tender and may be revised as projects are scoped. We have not asserted that vehicle-road-cloud integration is proven safer than car-only autonomy, because public evidence on that comparison is still emerging. ## What readers can do now - **For operators and suppliers:** map which of the twenty cities sit on your route or market, and track their roadside and terminal-fitting rules before any deployment. - **For researchers:** the pilot's emphasis on standards and test-evaluation is a signal that interoperability — not just vehicle intelligence — is where open technical questions remain. - **For observers:** check for an official mid-pilot progress report before concluding that infrastructure-assisted autonomy is scaling as planned. --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-policy-vehicle-road-cloud-pilot Free to quote with attribution and a link to the original.