Examples

Complete runs, each on a released dataset: the configuration used, the result produced, and — where a reference calibration exists — how far the two differ. Pick the one whose rig looks most like yours.

TUM VI

Rig: Stereo 1024² fisheye + BMI160 @ 200 Hz
Format: ROS1 bag
Shows: All six distortion models on identical data, vs a published Kalibr reference

Monado

Rig: Valve Index — stereo fisheye + IMU
Format: EuRoC mav0
Shows: Round-trip PASS and t_ga, vs the Monado reference baseline

Looper SA16

Rig: 2–3 cameras (one rolling shutter) + IMU @ 400 Hz
Format: Custom → EuRoC mav0
Shows: Four recording campaigns: rolling shutter, IMU oversampling, hardware sync

How to read these pages

Every example follows the same structure, so two of them can be compared directly:

The rig — sensors, rates, and what makes the dataset interesting.
Getting the data — download and any preparation the bag needs.
Configuration — the launch file, in full.
Running it — the command and what it prints.
Result — the calibration produced.
Validation — the same quantities from an independent source, and the difference. This is the part that makes an example evidence rather than a demonstration.
Reproducibility — code version, config, and data hash needed to obtain the same numbers.

Contributing an example

A run on a dataset we do not have is the most useful thing you can send. See Submit a dataset for the format, and Submit results if you want it on the leaderboard as well.