Obstructed by elevated road pillars, visual blind spots at urban intersections have long been a "hidden killer" of traffic accidents. In June 2026, China's first "Urban Traffic Risk AI Early Warning System," focused on addressing blind spots at elevated road intersections, was officially put into regular operation in Shanghai's Huangpu District. The system, jointly developed by Huangpu police and companies including Ping An Insurance (Group) Company of China, integrates cutting-edge technologies such as millimeter-wave radar, high-definition cameras, and edge computing to achieve real-time, precise perception of intersection dynamics. After two months of pilot operation, incidents of non-motor vehicles entering the elevated road area at the intersection have significantly decreased, and the traffic accident rate has notably dropped.

The "Hidden Killer" of Visual Blind Spots in Megacities

The AI traffic early warning system was deployed at the intersection of Huangpi South Road and Yan'an East Road, located in the core area of downtown Shanghai, adjacent to People's Square and the entrance/exit of the Yan'an Elevated Road. The area experiences dense pedestrian and vehicle traffic with complex traffic flow patterns. Due to obstruction by elevated road pillars, this intersection has fixed visual blind spots. Pedestrians and non-motor vehicle riders have obstructed views when passing through, leading to dangerous behaviors such as mistakenly entering restricted zones or cutting across lanes, which can easily cause casualties. This intersection has historically been prone to traffic accidents resulting in injuries and has been a key focus for traffic hazard remediation in the area.

In a megacity like Shanghai, intersections obstructed by similar elevated road pillars are not isolated cases. How to use technological means to solve the governance challenge of "seeing but not preventing" is a critical issue for the intelligent transformation of urban traffic management.

Triple Defense Line: Millimeter-Wave Radar + Edge Computing + Acoustic-Optical Warning

"Perception Layer": Millimeter-wave radar + high-definition cameras for all-weather, precise capture

The system integrates cutting-edge technologies such as millimeter-wave radar, high-definition cameras, and edge computing to achieve real-time, precise perception of intersection dynamics. Millimeter-wave radar is unaffected by weather conditions like rain, fog, or nighttime light, complementing high-definition cameras to enable all-weather, round-the-clock precise capture of pedestrians, non-motor vehicles, and motor vehicles within the blind spot area.

"Decision Layer": Deep integration of claims data and risk identification models

Huangpu police and Ping An have deeply integrated claims data, risk identification models, and intelligent perception hardware to innovatively create this urban traffic risk AI early warning system for visual blind spots in megacities. By cross-analyzing historical accident claims data with real-time perception data, the system can accurately identify high-risk behavior patterns and issue warnings before danger occurs.

"Intervention Layer": Ground guidance lights + dual acoustic-optical warnings for real-time "shouting" to avoid danger

The system projects guidance lights onto the ground to scientifically direct vehicle and pedestrian flows to detour in an orderly manner. Once it detects pedestrians or non-motor vehicles entering the pillar blind spot and engaging in dangerous behaviors such as crossing illegally, going against traffic, or cutting lanes, the system immediately triggers dual acoustic-optical warnings, with directional voice alerts reminding them to stay away from the dangerous blind spot area. The entire process from perception to warning is completed in milliseconds, achieving a real-time closed loop of "in-process intervention."

"Model Layer": Paradigm shift from "post-incident compensation" to "pre-incident prevention"

The deployment of this "Urban Traffic Risk AI Early Warning System" marks Ping An's transition from "traditional post-incident insurance compensation" to "pre-incident prevention and in-process intervention" for risk reduction and value enhancement. This is also a key step in Ping An's "Red Light" public welfare initiative, extending from "rural areas" to "urban areas" and from "basic protection" to "intelligent governance."

Governance Upgrade from "Single-Point Pilot" to "Full-Scale Promotion"

Significant pilot results provide a replicable model for megacity traffic governance

After two months of pilot operation, incidents of non-motor vehicles entering the elevated road area at the intersection have significantly decreased, and the traffic accident rate has notably dropped. This system has effectively improved the safety coefficient of urban road traffic, achieving a win-win situation in social, safety, and economic benefits. This achievement provides a replicable and scalable technical model for addressing blind spots at elevated road intersections in Shanghai and other megacities in China.

Insurance industry's "risk reduction" moves from compensation to prevention

Ping An stated that 2026 is the group's "Service Year." The deployment of this system is precisely Ping An's transition from "traditional post-incident insurance compensation" to "pre-incident prevention and in-process intervention" for risk reduction and value enhancement. Previously, in April 2025, Ping An launched the "Red Light" road safety risk reduction public welfare initiative, which has completed the renovation of 1,789 high-risk road sections across 31 provinces (autonomous regions and municipalities directly under the central government) and 303 counties in China, deploying over 10,000 sets of traffic safety facilities. The launch of this urban traffic risk AI early warning system marks the initiative's expansion from rural areas to urban areas and from basic protection to intelligent governance.

"Shanghai Experience" in smart urban traffic governance

In Shanghai, Ping An Property & Casualty Insurance has also implemented "one light, one strategy" smart lighting renovations for high-risk intersections with poor lighting and obstructed views, completing approvals and pilot planning for 23 key intersections; launched the "Rider Safety Co-building Plan," collaborating with Taobao Flash Delivery to provide customized safety training, traffic awareness campaigns, and accident insurance services, serving over 10,000 riders cumulatively; and partnered with traffic management departments to deploy intelligent early warning and dynamic duty mechanisms during peak holiday periods, achieving full-cycle safety assurance for three key scenic spots in 2026 with zero major accidents. This series of combined measures is building the "Shanghai Experience" in megacity traffic governance.

Scalable to more urban elevated road intersections

As technology matures and costs decrease, this system is expected to be promoted to more urban elevated road intersections, tunnel entrances, large overpasses, and other visual blind spot sections across China. When every "invisible corner" is "seen" by AI, the safety baseline of urban traffic will be redefined.

From the "Red Light" public welfare renovation of rural roads to the AI intelligent early warning at elevated road intersections in Shanghai's Huangpu District, Ping An and Huangpu police have used a technological combination of "millimeter-wave radar + edge computing + acoustic-optical warning" to explore a new path for megacity traffic governance, shifting from "post-incident handling" to "pre-incident prevention and in-process intervention." As Ping An has committed, the company will continue to uphold its original intention of "finance serving the people," collaborate with public security traffic management departments and various social forces, continuously explore more diversified risk reduction models, and strengthen traffic safety defenses with professional capabilities, technological power, and warm-hearted services. When AI "sees" blind spots and warnings "shout out" dangers, every trip in urban traffic will have an added layer of safety assurance.