# Ouster OS1-64 LiDAR Drives with Aptiv ESR Radar Tracks, Kalamazoo 2020-2025 (WMU AD Assurance Lab)

## Facts

- Provider: AD Assurance Lab, Western Michigan University
- License: CC-BY-4.0 (https://huggingface.co/datasets/AD-Assurance-Lab/ouster-lidar-drives)
- Size: 191 files
- Format: Parquet, JSON (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/AD-Assurance-Lab/ouster-lidar-drives
- Access: public, readable without an account
- Category: Robotics
- Source: https://huggingface.co/datasets/AD-Assurance-Lab/ouster-lidar-drives
- Page: https://scigantic.com/archive/5e70c4fd-0b2a-5d41-aa45-28fd08ade5a8
- Keywords: Robotics, LiDAR, Parquet, JSON, Hugging Face

## About

Real-world automotive LiDAR point clouds from the roof-mounted Ouster OS1-64 of the AD Assurance Lab at Western Michigan University, recorded in and around Kalamazoo, Michigan between 2020 and 2025: a 2020 highway drive, 2021 Kia Niro drives, a December 2023 urban drive, radar test drives in 2023-24, a January 2025 campus shuttle drive and camera-LiDAR calibration recordings. The 2020 and 2021 drives also carry Aptiv (Delphi) ESR 2.5 long-range radar tracks, so vehicles can be located independently of the LiDAR out to about 175 m, which makes the set useful for long-range object detection where a car returns only a few points. There are no 3D labels. Each lidar/scans_*.parquet row is one full 360-degree scan of 65,536 points; Ouster IMU, radar tracks, GNSS and per-drive JSON metadata are Parquet and JSON beside it, and the shuttle folder adds a registered PCD map. The card documents point ordering per driver version and sensor frames. About 33 GB in 191 files (measured 2026-11-21); read with pyarrow and numpy. Released under CC BY 4.0 by the lab; cite the paper listed for each drive.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("AD-Assurance-Lab/ouster-lidar-drives")
print(ds)
```

## Data structure

File types in sampled files: Parquet, JSON

Example file: `calibration_2023-24/2023-12-21_cam_lidar/lidar/scans_000.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/5e70c4fd-0b2a-5d41-aa45-28fd08ade5a8
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/5e70c4fd-0b2a-5d41-aa45-28fd08ade5a8
- Over MCP: `get_schema_card` with id `5e70c4fd-0b2a-5d41-aa45-28fd08ade5a8` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [JSON datasets](https://scigantic.com/topics/json)
- [LiDAR datasets](https://scigantic.com/topics/lidar)
- [Robotics datasets](https://scigantic.com/topics/robotics)
