Beijing Municipal Autonomous Vehicle Joint Working Mechanism β office of the Smart-City Infrastructure & Intelligent-Connected-Vehicle Co-Development Programme (εδΊ¬εΈεζΊε), delegated to the Beijing Economic-Technological Development Area administration β A municipally consolidated autonomous-vehicle data estate: the plan orders all operating data from every autonomous vehicle in the city to be uploaded into one citywide service-management platform, joined with vehicle-side device data, roadside-device data and city traffic data through the existing cloud-control base platform. Layered on that inβ¦
Beijing Municipal Autonomous Vehicle Joint Working Mechanism β office of the Smart-City Infrastructure & Intelligent-Connected-Vehicle Co-Development Programme (εδΊ¬εΈεζΊε), delegated to the Beijing Economic-Technological Development Area administration β A municipally consolidated autonomous-vehicle data estate: the plan orders all operating data from every autonomous vehicle in the city to be uploaded into one citywide service-management platform, joined with vehicle-side device data, roadside-device data and city traffic data through the existing cloud-control base platform. Layered on that inβ¦
Beijing offers (Alternative, Geospatial, Reference) β Order of several petabytes and rising. The programme's operating body has stated it has accumulated more than 7 PB of multi-source data β autonomous-vehicle supervision data, intersection panoramic perception data and high-definition mapping among it β and has already productised slices of it (a vehicle-to-everything dataset, a perception dataset, a pre-training dataset, an abnormal-event dataset and a hazardous-condition scenario library), with an earlier city-scale challenge-scene release citing roughly 800 autonomous vehicles and 300-plus instrumented intersections behind more than half a million labelled frames. The plan's contribution to volume is directional rather than numeric: it extends ingestion from the pioneer zone to the whole city and to three new industry verticals, so the base is a multi-petabyte estate being scaled citywide..
Infrastructure-side perception training β roadside camera and laser-radar streams with independent vehicle telemetry attached are the training substrate for the vehicle-to-everything and cooperative-perception model families the plan explicitly names as targets (multimodal large models, end-to-end driving, world models), and they arrive labelled by the vehicles' own uploaded state. World-model and simulation corpus construction from city-scale open-road scenario data, including the abnormal and takeover cases that mileage-based testing systematically misses. Regulatory-tier-conditioned safety benchmarking, comparing disengagement and takeover behaviour across the road-test, demonstration-application and pilot tiers and across bus, freight and passenger-car operations on identical roads. Signal-control and congestion modelling, since the platform fuses vehicle kinematics with the city's own signal and traffic data on the same road network. Insurance and risk pricing for driverless freight and passenger pilots keyed to the safety-assessment and inspection records rather than to operator claims. Competitive deployment intelligence for equity research from per-sector quotas and filed vehicle counts, which are set by the city before operators announce expansions. Map-and-corridor change detection, because the open-area register changes ahead of public service launches.
The data is with 6 years of history.
Coverage spans China/HK, US; Application Software, Passenger Ground Transportation, Trucking; alternative, geospatial, reference; equities.
Not yet on file: Summary & mission Β· Use cases & buyers Β· FAQ