search

cartographer-project / cartographer

Cartographer is a system that provides real-time simultaneous localization and mapping (SLAM) in 2D and 3D across multiple platforms and sensor configurations.
Apache License 2.0
7.09k stars 2.25k forks source link

IMU input data #1789

Closed dewtrullac closed 3 years ago

dewtrullac commented 3 years ago

Hi all,

I am using a 3DM-GX5-25 AHRS/IMU and 2 Velodyne VLP16 lidars, mounted in a similar way as in case of the Deutsches museum 3D public data. I have managed to figure out a lot (learn from READDOCS and github issues) about the Lidar and PointCloud2 msgs but I still have poor understainding of imu messages.

I am using the KumarRobotics imu_3dm_gx4 drivers. Since the 3DM-GX5-25 IMU also includes the Extended Kalman Filter at 500 Hz rate I enable the Kalman filter. However, my IMU msgs still remain the same. I then found out, that the filtered msgs are being published under another topic (/imu_filter as the FilterOutput msg type).

If I check individual imu topic msg I see that it is filled with linear acceleration and angular_velocity measurements of which covariance matrices are all zero. The orientation quaternion is always 0 and has the covariance matrix with the first element being -1 and others 0. 

  seq: 35552
  stamp: 
    secs: 1606133123
    nsecs:  16914899
  frame_id: "imu_link"
orientation: 
  x: 0.0
  y: 0.0
  z: 0.0
  w: 0.0
orientation_covariance: [-1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
angular_velocity: 
  x: 0.0706095993519
  y: 0.0323645025492
  z: -0.0257006008178
angular_velocity_covariance: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
linear_acceleration: 
  x: -1.69895464022
  y: -0.371063444654
  z: -9.55261377696
linear_acceleration_covariance: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]

By checking the imu_filter topic msg I see that there is orientation quaternion together with the filled covariance matrix.

header: 
  seq: 16086
  stamp: 
    secs: 1607437180
    nsecs: 351661636
  frame_id: "imu_link"
quat_status: 1
bias_status: 1
orientation: 
  x: -0.00177353795152
  y: -0.00374871050008
  z: 0.978669941425
  w: 0.205397248268
orientation_covariance: [1.4697543520014733e-06, 0.0, 0.0, 0.0, 1.4670102927993867e-06, 0.0, 0.0, 0.0, 5.634401168208569e-05]
bias: 
  x: -0.0002450463071
  y: 0.000240148568992
  z: -5.94623888901e-05
bias_covariance: [5.347048914750019e-10, 0.0, 0.0, 0.0, 5.346166287445442e-10, 0.0, 0.0, 0.0, 6.389778706150651e-10]
bias_covariance_status: 1
orientation_covariance_status: 1

Finally my question: Should my imu msgs also contain the orientation information and should the covariance matrices include some actual values (different than zero)? Or is it that the Poseextrapolator is doing what the IMU is doing with the EKF filtering and is in that case the IMU EKF filtering not necessary and not being used by the cartographer?

Kind regards, Dew

johuber commented 3 years ago

First of all you should spend some time understanding your sensors and what they do, it is not within our scope to help you with that and also not related to cartographer.

Short answer to your question: Only linear accelerations and angular velocities are used by cartographer. Orientation and covariance are not used.

dewtrullac commented 3 years ago

Thank you for your answer. Its what I wanted to know.