GI Ego1

Demo and values

One 30-second recording of an AprilGrid, solved with MVIS for camera, IMU and time offsets, with the rolling-shutter readout estimated too.

The 3D scene MVIS recovered from this recording. Everything except the board is blurred.

Reprojection median0.15 px

cam0–cam1 baseline62.1 mm

cam0 td+55.55 ms

cam0 lever arm31.9 mm

Reprojection is how far, in pixels, the detected corners lie from where the solve puts them. td is the camera–IMU time offset, and the lever arm is the camera–IMU position offset.

One recording, with no independent reference. These values show what MVIS solved, not how close they are to the truth.

Calibration process

The path of the rig around the board, as MVIS solved it. To explore it:

• Press Play or drag the bar to follow the path.
• Use the view buttons to change the view.
• Pick a camera card to put that camera on the frustum, the pyramid that shows its view.

Each card draws the board from the solved pose, with the corners MVIS detected. We publish no camera images.

Solved rig

The solved rig in 3D, to scale. Tap a sensor to see its values. In full screen you can rotate and zoom. Grid squares are 20 mm.

3DFrontTopSide+−

To scale, from above and the front

From above

From the front

Camera	Lever x, y, z (mm)	Distance (mm)	log(R) (rad)	td (ms)	Field of view
cam0	−12.7, −21.3, −20.2	31.9	[+0.023, −3.125, −0.021]	+55.55	~122° × 68°
cam1	+49.4, −20.4, −21.8	57.8	[+0.022, −3.117, −0.008]	+55.63	~128° × 68°

Solved by MVIS. Lever arms (camera–IMU position offsets) are in the IMU frame. Rotation is log(R), camera in the IMU frame.

Specifications

Measured in these recordings
Shutter	Rolling, readout about 26 ms (solved)
Cameras	2 × 1920×1080 at about 30 fps
Calibrated field of view	cam0 ~122° × 68° · cam1 ~128° × 68°
Published by the maker
Field of view	120° × 72°
IMU	200 Hz, 9-DoF
Not published
Open	Sensor part numbers

Dataset

Recording	2026-07-17, 30.1 s
Format	MCAP, H.265
Frames	Everything except the board is blurred
Frames kept at decode	cam0 889 of 900, cam1 about 859
Board	AprilGrid 6×6, tag 35.2 mm, spacing 0.3
Bags, configs and results	Calibench community, gi-ego1/ (access by automatic approval)
Licence	CC BY-NC 4.0, for the data and for this page's clip and figures; commercial use by request
Hosted	Calibench, gilabs-examples/ (access by automatic approval)

AprilGrid 6 × 6, 264 × 264 mm. Standard pattern, to scale

Detailed results

Lens Calibrated field of view

Calibrated field of view, horizontal × vertical. cam0: ~122° × 68°, corners reach 44° from the axis · cam1: ~128° × 68°, corners reach 50° from the axis

cam0 cam1

We read these from the calibrated lens fit at the image edges, through the principal point (where the optical axis meets the image). ~ marks an axis past where the corners reached, so the fit is extrapolated there. Past the fit means an edge lies beyond where the fit holds, so we give no value.

IMU Gyro–accel time offset t_ga −5.66 ms

Each bar shows a scale as its deviation from 1. IMU model: calyx. t_ga, the time offset between the gyroscope and accelerometer channels: −5.66 ± 0.64 ms.

Gyro scale x−3.27 %

Gyro scale y−4.02 %

Gyro scale z−3.46 %

Accel scale x+1.92 %

Accel scale y+2.45 %

Accel scale z+2.06 %

−8 %0+8 %

Timing cam0 td +55.55 ms

The time offsets, to scale. td is the camera–IMU time offset, with t_imu = t_cam + td.

Full results Every solved parameter

Rotation is log(R) in rad, camera in the IMU frame.

Parameter	MVIS
cam0
fx px	917.21
fy px	917.02
cx px	959.66
cy px	565.21
k1	−0.02889
k2	−0.00898
k3	−0.01316
k4	0.03078
Lever x mm	−12.67
Lever y mm	−21.28
Lever z mm	−20.16
log(R) rad	[+0.0228, −3.1248, −0.0213]
td ms	+55.545
Reprojection px	0.194
Readout ms	−26.48
cam1
fx px	922.97
fy px	922.72
cx px	894.01
cy px	561.58
k1	−0.03227
k2	−0.00100
k3	0.00497
k4	−0.01177
Lever x mm	+49.43
Lever y mm	−20.45
Lever z mm	−21.78
log(R) rad	[+0.0220, −3.1169, −0.0083]
td ms	+55.631
Reprojection px	0.198
Readout ms	−26.53
IMU
t_ga ms	−5.66 ± 0.64

Limits

• One recording, with no independent reference. These values show what MVIS solved, not how close they are to the truth.
• Gyro scale solves 3 to 4 % below 1 on all axes. We have not confirmed the cause.

How it was run

• Recording converted to EuRoC layout, camera-only solve per camera, then camera + IMU with camera intrinsics free, t_ga and the rolling-shutter readout both estimated. Batch of 2026-11-03.
• sha256 of the recording: f4443fc201d70088a28aa2011c1f82597ae911f2896745bdf267fa043f8aeae4