Connected Vehicle Cloud Control Platform

Definition and Technical Positioning

The connected vehicle cloud control platform, formally known as the intelligent and connected vehicle (ICV) cloud control infrastructure platform, serves as the core central hub of the vehicle-road-cloud integrated ecosystem. It is not merely an isolated cloud server, but a distributed cloud computing system built on unified standards, adopting a three-tier hierarchical architecture comprising the "edge cloud, regional cloud, and central cloud". Positioned as the "cloud brain" for intelligent connected vehicles, it aggregates massive data streams from vehicles, roadside infrastructure, and third-party platforms for sensor fusion, collaborative decision-making, and control. This enables foundational services such as beyond-line-of-sight perception, early hazard warnings, and traffic efficiency optimisation, serving as vital infrastructure backing the Chinese blueprint for "vehicle-road-cloud integration".

System Architecture: Three-Tier Cloud Collaboration

The technical architecture of the cloud control platform follows a hierarchical "edge cloud—regional cloud—central cloud" layout, with each tier operating collaboratively based on designated computing power and operational coverage. The Edge Cloud represents the layer closest to vehicles and road networks. It ingests real-time data from roadside radar, cameras, and on-board telematics units (TCUs/T-Boxes), leveraging lightweight local infrastructure to carry out low-latency data fusion and pre-processing. This delivers millisecond-level V2X collaborative services such as blind-spot warnings and cooperative vehicle control. The Regional Cloud consolidates data from multiple edge clouds to serve regional traffic authorities, offering near-real-time or offline services such as traffic flow analytics, traffic light optimisation, and fleet supervision, while managing in-depth analysis and long-term storage of cached edge data. The Central Cloud functions as the top tier, aggregating datasets across various regional clouds to supply macro-level traffic analytics and value-added raw data services to transport policy-makers, carmakers, and research bodies. This "layered and decoupled" setup relies on standardised tiered sharing interfaces to address varying latency demands from milliseconds to minutes across diverse scenarios—a defining trait setting it apart from conventional monolithic cloud platforms.

Core Technical Capabilities

The core capabilities of the platform are anchored in multi-source data fusion and distributed collaborative computing. At the perception layer, the platform collects roadside sensor feeds from cameras, millimetre-wave radars, and LiDAR units via edge gateways, alongside vehicle-uploaded CAN bus and on-board sensor data. It synchronises and fuses this information with third-party inputs like high-definition (HD) maps and weather data to generate structured trajectory tracking of road users. At the decision-making layer, built-in standardised service modules handle perception fusion, collaborative decision-making, and cooperative control, supporting connected autonomous driving use cases such as cloud-assisted Autonomous Emergency Braking (AEB), cloud-enabled Green Light Optimal Speed Advisory (GLOSA), and cooperative highway merging. Powered by the platform, a standard Level 2 (L2) test car can achieve seamless autonomous driving on public roads; roadside sensors spot hazards in visual blind spots early, allowing the cloud to dispatch braking commands directly to the vehicle and effectively overcoming the perception limits of standalone on-board sensor suites. Dongfeng Motor's in-house developed cloud control system consolidates parallel driving, T-Box connectivity, and V2X functions into a unified suite, capping end-to-end communication latency under 170 milliseconds.

Industry Deployment and Real-World Applications

Cloud control platforms have been rolled out across several "vehicle-road-cloud integration" pilot cities. In Beijing's Yizhuang 60-square-kilometre core zone, the platform supports large-scale deployment of smart infrastructure, validating use cases like cloud-backed Navigation on Autopilot (NOA). Nanjing is setting up a unified municipal-level intelligent connected cloud control platform using real-world 3D HD maps as a standardised spatial base layer. This tackles the alignment challenges of multi-source heterogeneous roadside data, realising end-to-end real-time monitoring of connected vehicles across the city. The Western China Smart Mobility Cloud Control Platform in Chongqing has been deployed across roughly 50% of national pilot cities, spearheading mass-production validation trials alongside 15 major domestic and global carmakers in the world's largest vehicle-road-cloud testing initiative. Meanwhile, Shenzhen's intelligent connected vehicle supervision platform leverages an end-to-end lifecycle monitoring framework to perform dynamic road risk evaluations and real-time tracking of operational vehicles.

Technical Limitations

The performance of cloud control platforms relies heavily on roadside sensor density, network stability, and the maturity of data fusion algorithms. Any sensor malfunction or connectivity drop can compromise cloud-based decision accuracy. In addition, platform standards across different regions are not yet fully harmonised, meaning cross-regional interoperability remains a work in progress. Given the heavy initial capital expenditure required for infrastructure rollout, establishing commercially viable and sustainable business models remains a key hurdle for the industry.