Why Ouster OS0-32 scores 77¶
Overall: 76.6/100 · Domain: SLAM · ROS 2
What bagx sees¶
The Ouster OS0-32 static capture is a useful teaching example: the LiDAR data is fine, but the IMU lane is marginal for LiDAR deskew.
| Metric | Value | bagx interpretation |
|---|---|---|
| IMU score | 79.6 | Acceptable noise, but not excellent |
| IMU rate | ~50 Hz | Below the 100–200 Hz sweet spot for deskew |
| Sync score | 73.6 | LiDAR↔IMU delay leaves little margin |
| Overall | 76.6 | Mean of IMU + sync (no GNSS in this bag) |
Why the score is not higher¶
- IMU frequency — At ~50 Hz, point-wise deskew competes with motion blur on faster rotations. bagx flags this relative to SLAM-oriented thresholds used for handheld LiDAR+IMU bags.
- Sync — A ~25 ms class delay between LiDAR and IMU erodes the sync sub-score even when timestamps look consistent.
- Short static segment — Allan bias metrics are limited on short recordings; bagx notes that longer static captures improve IMU characterization.
What to do before running SLAM¶
Typical mitigations:
- Increase IMU rate (hardware or driver config) if possible.
- Enable per-point deskew and verify sync with
bagx sync. - Record a longer static segment for IMU noise / bias estimation.
Reproduce¶
See the scoreboard for the full dataset list.