Calibrate an IMU against a sensor trajectory in your browser.

Estimates the rotation, clock offset, gyro bias, and lever arm between an IMU and a LiDAR (or any sensor with a TUM pose trajectory), with held-out evidence and a pass / warn / fail verdict. How the method is validated.

Runs locally via Pyodide; your data never leaves this page.

Have a ROS 2 bag? Plan calibrex check on it in the browser.

New here? Run the full pipeline on a generated 120 s hand-held recording with known truth, no files needed.

Or calibrate your own recording

IMU CSV t, gx, gy, gz, ax, ay, az — seconds, rad/s, m/s² or g Sensor trajectory (TUM) t x y z qx qy qz qw — poses T_world_sensor, e.g. LiDAR odometry Acceleration unit Livox ROS drivers report g

Compare with the Livox MID360 design extrinsic Rotation, clock offset, and gyro bias only (skip the lever arm)

Loading Python…

Results

Rotation, clock offset, gyro bias

Lever arm

Extrinsic

T_lidar_imu maps IMU-frame vectors into the trajectory's sensor frame; the clock offset is t_imu = t_sensor + dt.

The artifacts are schema-valid slac.imu_lidar_rotation/v0.1 and slac.imu_lidar_translation/v0.1 files with provenance; check them with calibrex validate.