TUM VI

Use this page to check a camera–IMU calibrator against a published answer. The TUM Visual-Inertial dataset is the reference check: a stereo-fisheye + IMU rig with a published Kalibr calibration to score against.

It is also the dataset where we compare every distortion model on identical data.

The rig

Sensor	Rate	Detail
2 × monochrome global-shutter cameras	20 Hz	1024×1024 fisheye
Bosch BMI160 IMU	200 Hz	/imu0

Board: AprilGrid 6×6, tag_size 0.088 m, tag_spacing 0.3, family t36h11.

Getting the data

You need two kinds of bag:

• Camera–IMU solve: the raw IMU-calibration bags, dataset-calib-imu{1,2,3,4}.bag.
• Camera-only fit: standard_1024/dataset-calib-cam1_1024_16.bag as the board bag, when a model needs a camera-only fit first.

tum_vi_calib/
  raw/dataset-calib-imu{1..4}.bag             # cam-IMU sequences
  raw/dataset-calib-cam1-camchain.yaml        # Kalibr omni/radtan intrinsics
  tumvi_camchain.yaml                         # Kalibr equidistant intrinsics
  standard_1024/dataset-calib-cam1_1024_16.bag
  omni-radtan-1024/ , pinhole-equi-1024/      # Kalibr cam-IMU reference results

One step, if a camchain exists

omni_radtan and equidistant ship with a Kalibr camchain for these cameras. A camchain is Kalibr’s YAML file of camera intrinsics and distortion. With it, the calibration is a single stereo run with the intrinsics pinned (held fixed):

mvis cam-imu -c examples/configs/tumvi_cam_imu/cam_imu_omni.yaml
mvis cam-imu -c examples/configs/tumvi_cam_imu/cam_imu_equi.yaml

Two steps, otherwise

double_sphere, eucm, and fov have no shipped camchain. Fit the camera first, then feed the result into the camera–IMU step:

mvis calibrate -c examples/configs/tumvi_cam_imu/mvis_cam0_ds.yaml
mvis cam-imu   -c examples/configs/tumvi_cam_imu/cam_imu_ds.yaml

Fit cam0 alone for the wide-FoV models

Stereo camera-only fits diverge on this dataset for double_sphere and eucm. They slide into the fx–xi / alpha basin, a fit that looks converged and is not.

The single-camera fit is stable. So the recipe fits cam0 only and runs a single-camera camera–IMU solve. See Camera model.

Results — imu1, 45 s window

The columns:

• Reproj: reprojection error, the pixel distance between detected and predicted corners.
• cam0 lever: the lever arm, the camera–IMU position offset of cam0.
• Baseline: the distance between the two cameras.
• td: the camera–IMU time offset.

Model	Mode	Reproj	cam0 lever (mm)	Baseline	td
omni_radtan	stereo, pinned	0.273 px	[45.0, −68.7, −43.2]	100.8 mm	−0.73 ms
equidistant	stereo, pinned	0.277 px	[44.7, −70.1, −43.7]	100.8 mm	−0.72 ms
double_sphere	cam0, pinned	0.275 px	[44.3, −68.5, −43.9]	—	−0.71 ms
eucm	cam0, pinned	0.275 px	[44.3, −68.3, −43.9]	—	−0.71 ms
fov	cam0, pinned	0.288 px	[44.0, −65.2, −44.0]	—	−0.68 ms
Kalibr (omni)	reference	0.11 px	[46.2, −70.9, −43.7]	100.94 mm	~0

What the table says

• Every model lands on the same physical rig. Lever ~[44, −68, −44] mm and td ~−0.7 ms across all six. Model choice changes how easily you get a good fit. It does not change the geometry a good fit recovers.
• Agreement with Kalibr: omni_radtan and equidistant reproduce the reference lever to ≤2 mm and the stereo baseline to ≤0.15 mm. The time offset is inside 1 ms.
• Across all four bags, omni_radtan holds: baseline 100.8–101.1 mm, lever within ~1 mm, and the freed IMU intrinsics (Dw/Da) consistent to under 1%. That cross-sequence spread is the honest error bar. It is better evidence than any single run’s residual.

Compare the parameters, not the residuals. MVIS’s reprojection error is higher than Kalibr’s 0.11 px because the two do not measure the same thing. They use different corner subsampling and robust statistics.

Pin the intrinsics in the camera–IMU step

Freeing them (free_intrinsics: true) lets the fx–xi (omni, ds) or alpha (eucm) gauge drift on a single sequence. It drifts along its equivalence class, a family of values that fit equally well, while reprojection stays flat.

Extrinsics and td stay correct, but the intrinsics are no longer uniquely pinned. Freeing the IMU intrinsics (imu_intrinsics: dw_da) is stable.

Reproducibility

The settings behind the imu1 results above:

Item	Value
Configs	examples/configs/tumvi_cam_imu/
Sequence	dataset-calib-imu1.bag, t0 3.0 s, 45 s window
Board	AprilGrid 6×6, 0.088 m, spacing 0.3, t36h11
Reference	Kalibr omni-radtan-1024/, pinhole-equi-1024/
Detection cache	/tmp/mvis_tumvi_detcache — re-runs repeat only the solve