OAK-D PoE stationary IMU does not have the gravitational constant as magnitude

RyanPaesschesooneVintecc

Hi,

I'am using an OAK-D PoE stereo camera on a mobile robot. The goal is to use a RTK GPS and the Oak IMU and magnetometer to predict it's position. However, when setting up the sensor fusion filter. I noticed that the norm of the IMU vector equals an average of 9.64 when the camera is stationary.

I am using the parameters below in a config file. This is given to the depthai-ros-driver

/oak_front:
  ros__parameters:
    camera:
      i_nn_type: none
      i_mx_id: 18443010E17A3C1300
    imu:
      i_enable_rotation: true
      i_message_type: IMU_WITH_MAG_SPLIT
      i_rotation_vector_type: ROTATION_VECTOR
      i_acc_freq: 100
      i_batch_report_threshold: 5
      i_gyro_freq: 100
      i_mag_freq: 80
      i_acc_cov: 10.0
      i_gyro_cov: 1.0
      i_magn_cov: 1.0
    stereo:
      i_subpixel: true
    rgb:
      i_fps: 5.0

A sample output of the IMU data:

header:
  stamp:
    sec: 1751357605
    nanosec: 756499822
  frame_id: oak_front_imu_frame
orientation:
  x: -0.66363525390625
  y: 0.249755859375
  z: 0.655029296875
  w: 0.26104736328125
orientation_covariance:
- -1.0
- 0.0
- 0.0
- 0.0
- -1.0
- 0.0
- 0.0
- 0.0
- -1.0
angular_velocity:
  x: -0.002130528911948204
  y: -0.004261057823896408
  z: -0.003195793367922306
angular_velocity_covariance:
- 1.0
- 0.0
- 0.0
- 0.0
- 1.0
- 0.0
- 0.0
- 0.0
- 1.0
linear_acceleration:
  x: -9.642200469970703
  y: -0.29742908477783203
  z: -0.20166100561618805
linear_acceleration_covariance:
- 10.0
- 0.0
- 0.0
- 0.0
- 10.0
- 0.0
- 0.0
- 0.0
- 10.0
---
header:
  stamp:
    sec: 1751357605
    nanosec: 765799822
  frame_id: oak_front_imu_frame
orientation:
  x: -0.66363525390625
  y: 0.249755859375
  z: 0.655029296875
  w: 0.26104736328125
orientation_covariance:
- -1.0
- 0.0
- 0.0
- 0.0
- -1.0
- 0.0
- 0.0
- 0.0
- -1.0
angular_velocity:
  x: 0.003195793367922306
  y: 0.008522115647792816
  z: -0.003195793367922306
angular_velocity_covariance:
- 1.0
- 0.0
- 0.0
- 0.0
- 1.0
- 0.0
- 0.0
- 0.0
- 1.0
linear_acceleration:
  x: -9.627717971801758
  y: -0.3165304362773895
  z: -0.2111855447292328
linear_acceleration_covariance:
- 10.0
- 0.0
- 0.0
- 0.0
- 10.0
- 0.0
- 0.0
- 0.0
- 10.0
---

Does anybody have an idea what's wrong. Should I calibrate the IMU?

jakaskerl

Hi RyanPaesschesooneVintecc
Yes. The IMUs don't come precalibrated so you would have to do it yourself to get proper norm reading. I've heard people succeeding with Kalibr, but perhaps you have some other tool you are more familiar with.

Thanks,
Jaka