At ZKBRAIN, a technology enterprise based in Haidian district, Beijing, an AI agent optimizes traffic signal timing at road intersections in real time, as displayed on its large demonstration screen.

In pilot zones, the agent can generate 200 sets of signal timing plans within 50 seconds to enable smarter traffic lights and smoother road traffic.

Leveraging 3D modeling, spatial intelligence and digital twin technologies, the agent creates digital replicas of on-road vehicles and pedestrians. AI cameras capture real-time traffic conditions, while proprietary algorithms enable precise traffic signal control, shifting the traffic management paradigm from vehicles adapting to fixed signals to traffic signals responding to real-time traffic flows.

Powered by the smart AI system, some problems in urban transportation have been effectively addressed.

For instance, traffic signals at intersections gain autonomous decision-making capabilities. Traditional traffic lights follow fixed, round-the-clock timing schedules, leaving vehicles to wait passively for signal changes. Such static control proves inefficient, causing wasted green-light time and prolonged vehicle queues during red-light cycles.

By monitoring real-time traffic flows through roadside sensors, the AI agent significantly reduces unnecessary vehicle waiting time and enhances overall road throughput. It autonomously adjusts green-light durations according to real-time traffic conditions; for example, green lights can be extended by one to 15 seconds during traffic congestion to meet actual vehicle passage demands. So far, the system has been deployed at 19 intersections across Haidian district.

To address the high supervision costs and excessive reliance on police manpower in traffic violation management, the AI agent provides real-time early warning and tracking for common traffic violations, including illegal lane changes by motor vehicles and red-light running by non-motor vehicles. This improves traffic management efficiency and reduces law enforcement costs.