Kaiyun Zhilian · Air-ground coordinated traffic solution

Build air-ground coordinationintegrated traffic solution

AI-centered capabilities for vehicle-road-cloud integration, intelligent traffic management, highway digital transformation, and low-altitude inspection — from roadside perception to a closed business loop.

Browse solutions

AI object recognition and behavior analysis in a traffic scene

Intelligent traffic perceptionViolation evidence, event recognition, and trajectory fusion

Vehicle-road-cloudPerception, analysis, and operational linkage

Multi-source perceptionVideo, radar, and trajectory fusion

Edge real-time inferenceFast roadside response

Scaled deliveryFits multiple traffic-governance scenes

Overall solution

One intelligence stack across four traffic scenes

Turn existing cameras on urban roads, expressways, and national/provincial highways into continuously running AI traffic-analysis nodes.

Vehicle-road-cloud integration

Use fusion-perception MEC as the roadside fulcrum to connect intersections, segments, and the cloud for real-time coordination and unified dispatch.

Intelligent traffic management

For violation evidence, event recognition, and intersection governance, form a loop that can be traced, linked, and reviewed.

Highway digital transformation

Build a small-model plus large-model warning agent to monitor road-network operations.

Low-altitude traffic inspection

Use air-ground analysis to automatically recognize key violations and abnormal events, and raise response efficiency.

Core product · KY-MEC-A2

Holographic fusion-perception MEC

Deploy video, millimeter-wave radar, and edge compute at the roadside to keep producing high-precision roadside perception data.

Multi-source fusionJoint processing of video, radar, and trajectories

Roadside real-time inferenceTraffic-event analysis on the device

Low-latency processingFast output of flow, speed, and queue length

Multi-scene fitFits intersections, arterials, and expressways

Calibrate video and radar automatically, and output flow, speed, queue length, and other traffic parameters.
Cover object structuring, trajectory fusion, and events such as lane-line crossing, wrong-way driving, and occupancy.
Support rule configuration, calibration, device registration, OTA, and platform interfaces.

Intersection governanceArterial coordinationExpressway perceptionEdge inference

Pilot-zone practiceThe Xiangyang C-V2X pilot zone has produced a working intersection fusion-perception MEC deployment.

Signal-control upgradeThe Pingshan, Shenzhen city-wide signal-control upgrade already supports multi-intersection coordinated governance.

Fusion accessA single device can carry coordinated perception from video and millimeter-wave radar.

Intelligent traffic governance

Urban, expressway, and highway traffic-governance equipment

Build a closed loop of violation recognition, standardized evidence, and AI pre-review. Reuse existing cameras, add intelligent analysis, and automatically identify and document violations.

Heavy trucks running red lightsRight turn without stoppingIllegal parking evidenceLane-line crossingFailure to yield to pedestriansWrong-way drivingReversingNot following lane guidanceEmergency-lane occupancyForced consecutive lane changesE-bike riders without helmetsIllegal e-bike passenger carrying

For urban roads, expressways, and national/provincial highways, covering 4 categories and 20+ violation and event types.

Intersection scene with intelligent recognition of traffic objects and violations

Highway digital transformation

Intelligent detection and warning for road-network operations

Small and large models work with agent orchestration so event warnings arrive sooner and with higher precision.

SenseMulti-source perceptionMulti-point, cross-scene video perception

UnderstandEdge understandingEdge small models analyze traffic operations in real time

CoordinateCentral collaborationCentral large models raise precision in a second pass

CloseBusiness closed loopAgents close the loop of review, reporting, and handling

Controllable data security

Train, test, and deploy models locally to reduce the risk of sensitive traffic data leaving the site.

Small and large model collaboration

Edge small models handle high-frequency events in real time, while central large models judge complex scenes and improve recognition precision.

Continuous algorithm iteration

For long-tail events such as debris, accidents, and fire smoke, algorithms can be built quickly and improved against on-site operations.

Low-altitude inspection · air-ground coordination

Air-ground coordination for a new traffic-governance model

Docks, drones, and an AI analysis appliance work together to inspect key bridges, road segments, and congestion points, connecting discovery, on-site arrival, and handling feedback into one loop.

Automatic route generationFlight-status monitoringAI capture and analysisAbnormal-event warningViolation evidence captureRemote loudspeaker handling

Drones and docking stations inspecting roads together

Typical applications

Reusable traffic capabilities from demonstration projects

The solution is already in use in vehicle-network pilot zones, signalized-intersection upgrades, and road-governance programs.

C-V2X pilot zoneIntersection fusion perception

The Xiangyang C-V2X pilot zone has produced a working intersection fusion-perception MEC deployment.

Signal-control upgradeConnected coordination

The Pingshan, Shenzhen city-wide signal-control upgrade already supports multi-intersection coordinated governance.

Traffic governanceMulti-scenario delivery

The solution covers expressways, urban roads, and national/provincial highways, and is already live in dozens of cities.

Solution inquiry

Build a traffic system that can sense, coordinate, and decide.

Get the standard proposal, product parameters, and scene-specific recommendations.