Aerial Dataset for China Congested Highway & Expressway

AD4CHE

A drone-based aerial naturalistic driving dataset capturing congested traffic on Chinese highway expressways — rich car-following, lane-change, and forced-merge interactions under constrained traffic flow.

Raw drone footage
Standardized trajectories + HD map --
Loading video…
Loading map…
00:00 / 00:00

Drag the slider to compare raw drone video with standardized trajectories on the HD map — sample case from AD4CHE.

Dataset Overview

AD4CHE focuses on the safety-critical part of expressway driving — congested traffic — where safety-critical interactions concentrate and conventional in-vehicle datasets struggle to capture clean trajectories.

Naturalistic Trajectories

High-precision vehicle trajectories captured from a top-down drone perspective. Dense car-following, frequent lane changes, and forced merges that ground-level sensors miss.

Congestion-Focused

Recordings target stop-and-go waves, short-headway intrusions, and constrained-gap merging — the long-tail traffic states where most safety-critical interactions occur.

Adopted Worldwide

Used by 400+ academic and industrial institutions for ADAS calibration, motion-prediction benchmarks, scenario extraction, and human-like driving model training.

Key Numbers

Confirmed metrics from the AD4CHE release. Refer to the IEEE T-IV paper for the full breakdown by site and recording.

400+
Institutions Applied
4K
Aerial Resolution
25 Hz
Trajectory Sampling
2023
First Released

What's Included

Each release ships with rich kinematic and contextual fields, ready for direct scenario analysis.

Per-Trajectory Fields

  • Position (x, y) & velocity (vx, vy)
  • Longitudinal & lateral acceleration
  • Heading angle & yaw rate
  • Vehicle length, width, type
  • Persistent track ID across frames
  • Per-frame timestamps at 25 Hz
  • Lane ID & lane-relative position
  • Surrounding vehicle IDs (front / rear / left / right)

Safety & Scenario Tags

  • Distance headway (DHW)
  • Time headway (THW)
  • Time-to-collision (TTC) & minTTC
  • Lane-change events
  • Cut-in / cut-out flags
  • Mainline vs. ramp tagging
  • Per-recording metadata file
  • Documentation & sample loader scripts

How to Cite

Please cite the AD4CHE paper in any publication that uses this dataset.

"The AD4CHE Dataset and Its Application in Typical Congestion Scenarios of Traffic Jam Pilot Systems"

Published in IEEE Transactions on Intelligent Vehicles · Jilin University AD Safety Joint Lab · Zhuoyu Technology

Application for Access

Free for non-commercial academic research. We review every request manually and typically respond within 5 business days.

Application is submitted via our Feishu form. We typically respond within 5 business days.
No Feishu account required — choose "Fill in as guest" on the login screen if you don't have one.

Open Application Form →

Other Open Datasets

Companion open datasets from the lab and partner institutions, covering different road types and traffic agents.