China’s transportation authorities are turning the broad idea of “AI plus transportation” into a large list of practical tests across roads, railways, waterways, aviation, ports, and postal networks. A national action plan issued in June by the Ministry of Transport and four other departments has identified 860 pioneering application scenarios and set a 2030 ambition for new AI-driven infrastructure, equipment, service models, and industry forms. The policy’s scale became clearer in a detailed People’s Daily report published September 1, which describes a first batch of 41 typical scenarios and examples already operating around the country.
The story is not that every transport system is becoming autonomous at once. China’s plan is a framework for deployment, with different applications at very different levels of maturity. Some are familiar operational tools, such as traffic alerts, automated sorting, or machine-vision safety monitoring. Others, including autonomous heavy trucks and more sophisticated multimodal coordination, face harder engineering, safety, regulatory, and commercial questions.
That distinction matters. A large inventory of scenarios is not equivalent to an equally large fleet of deployed systems. But it does show how Chinese agencies are trying to create a common route from AI development to real-world infrastructure projects, which can give technology providers places to test systems, gather feedback, and build paying use cases.
A Map of 860 Application Scenarios
The Ministry of Transport has identified the scenarios across railways, highways, waterways, civil aviation, and postal services. The plan’s priorities stretch from AI-supported driving and roadside systems to rail operations, civil aviation, parcel networks, and connected modes of freight and passenger travel. Rather than concentrating only on passenger cars, the approach covers the full physical network that moves people and goods.
The first 41 guidelines are intended to make typical applications more replicable. That is an important policy function. Local authorities and companies often have pilot projects, but moving from a site-specific demonstration to a widely used operating model requires technical standards, procurement rules, safety practices, and a clear picture of who bears responsibility when the technology fails.
The examples in the People’s Daily report range from passenger-flow management and route optimization to fog warnings on expressways, security checks, and cold-chain monitoring. The breadth reflects the fact that transportation is a data-rich but highly fragmented sector. A port, a railway tunnel, a city bus network, and a last-mile delivery service all have different equipment, operating conditions, and risk profiles. A single general-purpose model cannot simply be placed on top of every one.
The policy is also aimed at public outcomes as well as industrial development. Better route information, earlier warnings, and more reliable sorting can improve convenience and resilience without making a headline-grabbing autonomous vehicle the focal point. That may be why the plan includes both visible deployments and back-office systems that passengers rarely notice.
Where Deployment Is Already Visible
Several of the report’s examples describe operational systems rather than future concepts. On the Beijing-Xiong’an Expressway, Hebei Expressway Group says it uses Baidu AI Cloud and large models to provide rapid alerts and one-click notifications for abnormal incidents. The company said its system identifies emergencies with more than 95 percent accuracy, a performance claim that should be understood as the operator’s own account.
The postal and express-delivery network offers more concrete scale. According to the People’s Daily report, more than 16,000 autonomous delivery vehicles and more than 400 drones were operating across China’s sector. The same report says AI is being used for route planning, sorting, and last-mile operations. These are not all identical services. Some delivery vehicles operate on constrained routes, while drones and automated sorting machinery require their own rules for airspace, facilities, and human supervision.
Railway safety is another major testing ground. China’s high-speed-rail network has linked more than 56,000 cameras into monitoring systems covering locations such as tunnel entrances and bridge approaches, the report says. Combining video and radar data, the systems are designed to identify people or foreign objects on tracks.
By April 2026, the network had processed close to 310,000 alerts judged to be valid. The figures illustrate why infrastructure AI is often more about monitoring and response than about full autonomy.
This policy direction connects with a trend EastFrontier examined in July, when China’s autonomous last-mile delivery vehicles were expanding rapidly. The national transport plan gives that kind of commercial deployment a broader public-policy setting, while also extending the agenda well beyond delivery.
From Pilots to an Operating System
The key challenge is not finding possible uses for AI. It is deciding which uses can operate safely and economically across varied regions and operators. An algorithm that works at one automated port may need extensive adjustment before it fits another. A roadside warning system may require reliable sensor maintenance, network coverage, and procedures for human staff. Scaling depends on operational discipline as much as model capability.
The action plan’s 2030 horizon suggests that officials recognize this will be gradual. The goal is to establish an ecosystem in which infrastructure, equipment, service providers, carriers, local governments, and AI companies can develop applications together. It is not a timetable for replacing people throughout the transport system.
Autonomous driving will remain one high-profile component, but it is only one. EastFrontier’s August coverage of China’s road-safety-law rewrite showed that legal rules are evolving alongside technology. The transport plan adds another layer: a portfolio of use cases where authorities can turn AI from a general aspiration into specific infrastructure and service projects.
For Chinese AI firms, transport can provide valuable real-world environments that are difficult to replicate in a lab. For transport operators, the technology offers potential gains in safety, reliability, and cost control. But both sides still have to prove that systems work in rain, fog, congested facilities, and complex human environments. China’s 860-scenario map is therefore best understood as an organized attempt to test that proposition at national scale.