LooperRobotics Insight 9

Demo and values

ChessboardRobot arm 1Robot arm 2

Sensor head from Looper Robotics with two global-shutter mono cameras, one rolling-shutter colour camera and a 400 Hz IMU. Our own unit on a chessboard, and the hand-held and robot-arm recordings, are shown separately.

The 3D scene MVIS recovered from chessboard take 04 of our own unit. Everything except the board is blurred.

Reprojection median0.14 px

cam0–cam1 baseline99.5 mm

cam0 td+1.81 ms

cam0 lever arm45.4 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.

No reference exists for these recordings, so the values can't be checked for accuracy.

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	−11.9, +39.8, +18.4	45.4	[+0.006, +0.003, −1.574]	+1.81	96° × 113°
cam1	−11.8, −59.8, +17.6	63.4	[+0.004, +0.001, −1.580]	+1.79	96° × 113°
cam2	−11.8, −9.9, +17.4	23.3	[+0.002, +0.009, −3.140]	−5.45	~149° × 82°

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

Specifications

Published by the maker
Type	Spatial AI camera
From our catalog
Global-shutter cameras	2 × 544×640 mono at 20 Hz, hardware-synced stereo
Rolling-shutter camera	1 × 1088×1920 colour at about 4.3 Hz
Lens	Wide field of view; pinhole-equidistant fit
IMU	1 × 400 Hz, with stamps quantized to 2.5 ms
Measured in these recordings
Calibrated field of view	cam0 96° × 113° · cam1 96° × 113° · cam2 ~149° × 82°
Not published
Open	Sensor part numbers

Dataset

Chessboard	4 recordings of 36 to 61 s, our own unit: takes 01 to 03 warm (41 to 43 °C), take 04 at 55 °C. This page shows take 04
Chessboard data	Calibench community, looper-insight9/data/ (access by automatic approval), takes 01 to 04 (this page shows take 04). Everything except the board is blurred. CC BY-NC 4.0, for the data, clip and figures; commercial use by request
Hand-held	10 of about 66 s, 15.0 GB in all
Hand-held data	Calibench, looper-insight9/ (access by automatic approval). CC BY-NC 4.0, for the hand-held data, clip and figures; commercial use by request
Robot arm	2 of about 58 s, same program in both
Robot-arm data	Calibench community, looper-insight9/data/ (access by automatic approval). Normal frames. CC BY-NC 4.0, for the data, clip and figures; commercial use by request
Board	Chessboard: 10×7 squares (9×6 inner corners), 43.47 mm, shown on a monitor. Hand-held and robot arm: AprilGrid 6×6, tag 55 mm, spacing 0.3, t36h11
Configs and results	Calibench community, looper-insight9/ (access by automatic approval)
Worked example	Looper SA16

AprilGrid 6 × 6, 412 × 412 mm. Standard pattern, to scale

Detailed results

Lens Calibrated field of view

Calibrated field of view, horizontal × vertical. cam0: 96° × 113°, corners reach 58° from the axis · cam1: 96° × 113°, corners reach 57° from the axis · cam2: ~149° × 82°, corners reach 60° from the axis

cam0 cam1 cam2

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 −5.43 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: −5.43 ± 0.03 ms.

Gyro scale x−0.30 %

Gyro scale y+0.37 %

Gyro scale z−0.36 %

Accel scale x−1.03 %

Accel scale y−0.01 %

Accel scale z−0.22 %

−8 %0+8 %

Timing cam0 td +1.81 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	330.82
fy px	330.65
cx px	270.35
cy px	320.15
k1	−0.02366
k2	0.01731
k3	−0.02395
k4	0.00895
Lever x mm	−11.86
Lever y mm	+39.77
Lever z mm	+18.38
log(R) rad	[+0.0056, +0.0030, −1.5745]
td ms	+1.812
Reprojection px	0.156
cam1
fx px	330.52
fy px	330.42
cx px	269.76
cy px	321.15
k1	−0.02632
k2	0.02520
k3	−0.03435
k4	0.01329
Lever x mm	−11.80
Lever y mm	−59.75
Lever z mm	+17.62
log(R) rad	[+0.0042, +0.0010, −1.5799]
td ms	+1.792
Reprojection px	0.163
cam2
fx px	769.74
fy px	769.82
cx px	959.50
cy px	538.16
k1	−0.03519
k2	0.01102
k3	−0.01813
k4	0.00942
Lever x mm	−11.82
Lever y mm	−9.93
Lever z mm	+17.41
log(R) rad	[+0.0018, +0.0086, −3.1403]
td ms	−5.445
Reprojection px	1.782
Readout ms	+31.63
IMU
t_ga ms	−5.43 ± 0.03

Limits

• No reference exists for these recordings, so the values can't be checked for accuracy.
• Hand-held and robot-arm recordings are kept apart, not pooled. They were solved in different batches and may not come from the same unit.
• The arm program rotates mostly about one axis, which leaves lever arms less well determined, so treat the arm's values with more caution.
• The chessboard recordings come from our own unit, not the partner's, and are shown on their own. The values shown come from take 04 alone, so they show no repeatability; all 4 takes are in the hosted data.

How it was run

• Camera + IMU with camera intrinsics free, t_ga and the cam2 rolling-shutter readout both estimated. Chessboard recordings of 2026-11-07; hand-held batch of 2026-11-02; robot-arm batch of 2026-11-04.