When vehicles are no longer isolated islands that merely "act according to traffic lights," but can instead "see" pedestrians in blind spots ahead, understand ambulance yield requests, and even find their own parking spaces after the driver leaves—this kind of mobility is moving from the test roads in Beijing Yizhuang to mass-production shelves.
On October 23, the 2026 World Intelligent Connected Vehicles Conference concluded successfully at the Beiren Yichuang International Exhibition Center in Beijing Yizhuang. During the conference, Beijing Vehicle-Network Technology Co., Ltd. (Beijing VNET), in collaboration with Academician Li Keqiang's team from Tsinghua University, partnered with 22 domestic and foreign automakers to launch four new connected vehicle functions for real-vehicle applications. Simultaneously, the company, together with YunKong ZhiXing, signed mass-production application agreements for Vehicle-Road-Cloud Integration with FAW, Toyota, and BAIC. This marks the industry's initial formation of a commercial closed loop comprising "technology verification - functional services - mass production application," declaring that "Vehicle-Road-Cloud Integration" functional services have officially "opened for business."
As early as July 2024, five ministries launched pilot programs for Vehicle-Road-Cloud Integration applications. In September this year, nine departments including the Ministry of Industry and Information Technology (MIIT) jointly issued the "15th Five-Year Plan for the Development of the Intelligent Connected New Energy Vehicle Industry," explicitly proposing to promote the distribution of traffic information to the vehicle side and accelerate the large-scale implementation of V2X scenarios. Under this policy tailwind, the newly added series of connected functions effectively validated the practical advantages of Vehicle-Road-Cloud Integration in urban travel safety assurance, efficiency improvement, and emergency response.
Three Functions Launched: "Blind Spot Safety" and "Urban Emergency Response" in Real-World Tests
Among the newly added functions, each directly addresses real pain points in urban mobility.
Connected Vulnerable Road User Collision Avoidance (C-VRU) focuses on vulnerable road users located in blind spots. Test data shows that when an electric two-wheeler ignores signal lights and makes a high-speed left turn across an intersection, the test vehicle immediately receives a cloud-based warning, calmly completing deceleration and avoidance when there is still approximately 4 seconds of driving distance remaining before reaching the two-wheeler. When roadside monitoring detects occluded pedestrian sightlines and transmits this information to the cloud in real-time, the cloud platform assesses the situation within 50 milliseconds and alerts the vehicle to "ghost probe" risks. The vehicle then completes a smooth brake before the pedestrian passes. This makes "invisible dangers" perceivable and avoidable.
Connected Automated Valet Parking Guidance (C-AVP) tackles the century-old problem of urban parking. Relying on the cloud control platform to aggregate parking lot platform data, it enables global path planning while the owner is away from the vehicle, synchronizes driving status in real-time, and autonomously achieves spot finding, parking in, parking out, and remote summoning responses. Tests show that compared to single-vehicle AVP, C-AVP allows 10 vehicles to park simultaneously in coordination, saving over 8% in average parking time and significantly reducing travel distance—the "multiplicative effect" of parking efficiency becomes evident.
Special Vehicle Warning addresses challenges such as limited response times and numerous interference factors in traditional warning modes by utilizing the cloud control platform to coordinate warnings and yielding among connected vehicles. During testing, ambulances executing tasks report emergency events instantly via onboard terminals. The cloud control platform identifies and predicts the driving status of all connected vehicles within a 1-kilometer radius along the ambulance's path within 100 milliseconds, issuing reasonable lane-change suggestions. Relevant vehicles complete deceleration or lane changes within 3-5 seconds, securing a precious "green channel" for life-saving emergencies.
Ecosystem Expansion: New Automakers Join, Mass Production Closed Loop Takes Shape
To bridge the gap between technology verification and commercial closed loops for Vehicle-Road-Cloud Integration, Beijing VNET collaborated with Academician Li Keqiang's team from Tsinghua University as early as 2025, partnering with 15 domestic and foreign automakers to achieve real-vehicle validation and mass-production planning for two services: Cloud-Supported Green Light Speed Advisory (C-GLOSA) and Cloud-Supported Automatic Emergency Braking (C-AEB). In 2026, industry participation further increased, with six mainstream automakers including Xiaomi Auto and Seres Auto joining. Deployment areas expanded to include Boxing Road, Rongxing North Third Street, and the National Intelligent Connected Vehicle Innovation Center.
The "circle of friends" for industrial collaboration is expanding rapidly. According to Academician Li Keqiang, a Member of the Chinese Academy of Engineering, Professor at Tsinghua University, and Chief Scientist of the National Intelligent Connected Vehicle Innovation Center, who spoke at the conference, the number of domestic and foreign companies participating in the consistent action plan for Vehicle-Road-Cloud Integration has expanded from 15 to 22, with 21 vehicle models participating in mass-production development and testing activities. These include independent automakers such as FAW, Dongfeng Motor, Changan Automobile, BYD, Xiaomi, and NIO, as well as foreign automakers like Toyota. Models such as the FAW Tiangong 07, BYD Han, Xiaomi YU7, and Toyota RAV4 are among them.
This mass-production signing represents a critical step for Vehicle-Road-Cloud Integration moving from technology verification to mass-production application. Beijing VNET, in conjunction with YunKong ZhiXing, signed agreements with FAW, Toyota, and BAIC focusing on high-frequency essential scenarios involving environmental perception at over 300 intersections and road construction. Leveraging their respective unique advantages in developing information resources and safety management for Vehicle-Road-Cloud Integration, as well as scenario innovation and industrial collaboration, they help automakers ensure secure access and real-time interaction of traffic information. Additionally, they assist in building accompanying cloud platforms and connected function services, providing continuous support for subsequent mass-production applications.
Market and Industry: Initial Supply Chain Formation, Trillion-Yuan Blue Ocean Opens
This series of actions resonates highly with the overall trajectory of the industry.
Academician Li Keqiang stated at the conference that the Vehicle-Road-Cloud Integration supply chain has initially formed, and commercial contracts between upstream and downstream players have gradually begun. To date, cooperation has been established with 14 groups of upstream and downstream enterprises, totaling 26 industrial partners. These include 11 OEMs, 8 city vehicle-network companies, and 7 suppliers of new types of connected components, spanning provinces and cities such as Beijing, Shanghai, and Chongqing. As of July 2026, intelligent roadside infrastructure covers approximately 32,000 kilometers of roads, with the scale of intelligent connected intersections continuously expanding. Li emphasized: "Vehicle-Road-Cloud Integration is not a technical route opposed to single-vehicle intelligence, but rather a direction of technological evolution for the further development of single-vehicle intelligence." In his view, breaking through the limitations of single-vehicle intelligence and building reliable infrastructure are the two keys to forming industrialized system capabilities.
The market space has also been repeatedly confirmed. According to predictions by EqualOcean Intelligence, China's V2X market size is expected to reach CNY 496 billion by 2030. The National Intelligent Connected Vehicle Innovation Center predicts that by 2030, the incremental total output value of China's Vehicle-Road-Cloud Integration intelligent connected vehicle industry will reach CNY 2.5825 trillion, with a compound annual growth rate of 28.8%.
On the policy front, the "15th Five-Year Plan for the Development of the Intelligent Connected New Energy Vehicle Industry" specifies that by 2030, NEVs will account for 70% of new car sales, and vehicles equipped with autonomous driving functions will achieve large-scale application. It specifically deploys measures to promote the development and application of Vehicle-Road-Cloud Integration, driving digital and connected upgrades of roads in key cities. Industrial clustering effects are accelerating in the Beijing-Tianjin-Hebei region—as of early 2026, the region has fully opened 4,311 kilometers of autonomous driving test roads, accumulating over 60 million kilometers of test mileage, making it the largest cross-regional intelligent connected vehicle testing scenario in the country by volume.
Background: From "Information Interconnection" to "User-Friendly Onboard Products"
Vehicle-Road-Cloud Integration did not emerge out of nowhere; it began with a technical exploration of "how to balance efficiency and safety," and key milestones have occurred in Beijing Yizhuang.
In the Beijing Dual-Intelligence City Innovation Demonstration Zone, Beijing VNET, the dedicated operating platform, had already collaborated with Academician Li Keqiang's team in2025. Partnering with mainstream automakers such as FAW, Dongfeng, Geely, NIO, GAC, SAIC-GM-Wuling, BAIC, and SAIC, they were the first to complete onboard testing and verification of two core functions: C-AEB and C-GLOSA. The successful validation of these two functions opened the door to intelligent mobility and led the industry to recognize that Vehicle-Road-Cloud Integration is no longer just a technical solution, but a foundational technical architecture capable of delivering specific, user-friendly products.
Real-world tests in Beijing Yizhuang wrote the fatal scenario of the "disappearing lead vehicle" into textbooks. On urban expressways, vehicles follow the lead car smoothly, with visibility completely obstructed. Suddenly, the lead vehicle changes lanes, revealing a stationary vehicle due to a breakdown—a common breeding ground for rear-end collisions. In single-vehicle intelligence mode, both human drivers and NOA systems relying on onboard sensors may cause accidents due to insufficient reaction time. However, in this real-world test, when the ego vehicle was still obscured by the lead car, the C-AEB system intervened: roadside sensing devices captured risks in advance like a "sky eye," the cloud control platform completed assessment and decision-making instantly, and issued early warnings and braking commands to the ego vehicle via the V2X channel, ultimately achieving safe avoidance. Test results showed that C-AEB secured a precious "golden second" for the vehicle, effectively reducing the last 5% of long-tail risks.
C-GLOSA turned the vision of "green lights all the way" into a command. On a 3-kilometer urban road containing 7 signal-controlled intersections, vehicles traditionally could only passively "act according to traffic lights." Once C-GLOSA was activated, roadside equipment uploaded real-time signal light data and intersection queue conditions to the cloud. The cloud control platform, acting like an experienced "urban traffic brain," synthesized global information to issue optimal speed recommendations. In 12 rigorous tests, applying C-GLOSA increased the number of non-stop passes throughout the trip to 3 times, a 200% improvement over single-vehicle intelligence mode; the number of passes with only one stop reached 9 times, a 125% increase, while expected average energy consumption decreased by at least 5%.
Technical Core: The Cloud's "God's Eye View" and "Triple Protection"
The technical core of these two functions is precisely what distinguishes Vehicle-Road-Cloud Integration from single-vehicle intelligence.
Take C-AEB as an example: it constructs a triple protection system of "roadside sensing - cloud assistance - vehicle execution." High-precision sensing units deployed on the roadside scan the road uninterruptedly 24/7. Sensing latency is ≤200 milliseconds, comprehensive positioning accuracy is ≤0.75 meters, speed accuracy is ≤0.8 m/s, and heading angle accuracy is ≤2.5°, acting as tireless "road domain sentinels." The cloud control platform aggregates multi-source information for global risk assessment, not only "seeing" dangers invisible to the ego vehicle but also predicting conflicts in advance. The vehicle side, via its onboard terminal, receives collision warnings and braking commands based on unified vehicle-cloud protocols, arbitrates with its own decision-making system, and executes the optimal braking strategy.
Experts from Tsinghua University's State Key Laboratory of Automotive and Traffic Engineering commented incisively: "Unlike the extreme braking of single-vehicle AEB, C-AEB can detect risks in advance, achieving significant advantages of strong predictive capability, lower braking intensity, and better speed smoothness." This "1-second head start" golden window is the irreplaceable value of Vehicle-Road-Cloud Integration in enhancing safety redundancy. In all test scenarios, participating vehicles achieved 100% successful obstacle avoidance—it does not replace single-vehicle AEB but serves as a powerful "cloud seatbelt," upgrading the safety system from "passive protection" dependent on single-vehicle hardware to "active warning and collaborative protection" integrating infrastructure.
C-GLOSA breaks two major shortcomings of traditional navigation. First, direct data connection: the cloud control platform connects directly with traffic signal control systems, obtaining real-time light status information with millisecond-level latency. Second, full-domain perception: roadside cameras and radars capture all elements such as vehicle queues in various directions and pedestrian crossings at intersections. The cloud control platform performs global optimization calculations not for a single vehicle, but for the entire traffic flow.
As stated by the relevant person in charge of Beijing VNET's Vehicle-Road-Cloud Integration project: "Under C-GLOSA technology, the calculated speed recommendations are smarter, more precise, and better aligned with actual traffic conditions, thereby truly improving the driving experience and efficiency." Tests also showed that signal light data upload latency is controlled within 200 milliseconds, with accuracy exceeding 99%, laying a solid technical foundation for future city-scale deployment.
Future Outlook: From Technical Leadership to Large-Scale Operations
The significance of the mass-production signings goes far beyond three agreements.
In terms of mass-production pace, Academician Li Keqiang provided a clear timeline: mass-production functions are expected to be onboarded via OEM infotainment APP/APIs, "official app stores/shelves," and infotainment driving interaction interfaces within this year. Integration via combined infotainment and intelligent driving systems is scheduled for completion by 2027. This means users will soon pick up cars from 4S dealerships that are truly "pre-installed" with Vehicle-Road-Cloud capabilities, rather than demo samples confined to test grounds.
From an industrial collaboration perspective, with the unification of industry standards and the entry of more automakers and cities, Vehicle-Road-Cloud Integration is poised to become a high-tier feature configuration for intelligent connected vehicles, spawning entirely new supply chains for high-precision roadside sensing units, cloud control platform application development, and V2X security services.
The next steps outline four paths: adhere to the coordinated development route of Vehicle-Road-Cloud Integration to consolidate the foundation of the Chinese solution; promote collaborative development and testing actions to continuously expand mass-production scale; strengthen high-quality urban data construction and operations maintenance to solidify the development base; and extend highway automaker collaborative development actions to achieve inter-city connectivity.
Conclusion
Interconnecting information merely opened the door to intelligent mobility; mass-production signings accelerate the world behind that door becoming daily reality.
Starting from the test roads in Beijing Yizhuang, C-AEB and C-GLOSA allowed us to see achievable "safety" and "efficiency," while C-VRU, C-AVP, and Special Vehicle Warning extended this capability to address urgent needs regarding travel blind spots, parking, and urban emergency response. Vehicles are no longer information silos but harmonious mobile nodes within smart cities.
As the core construction and operation platform for the "Dual-Intelligence" city, Beijing VNET will continue to drive China's Vehicle-Road-Cloud Integration from technical leadership to large-scale operations, making future mobility safer, more efficient, greener, and more convenient. Although the journey ahead is long and arduous, the industry has found its direction and taken a solid step forward.