Items with no label
3335 Discussions

questions about imu information of ZR300, how to calibrate the raw imu data

蔡之奡
Beginner
1,653 Views

I am trying to use ZR300 to run an opensource VIO, VINS-Mono(https://github.com/HKUST-Aerial-Robotics/VINS-Mono https://github.com/HKUST-Aerial-Robotics/VINS-Mono). I find that the output of the imu date topic:/camera/imu/data_raw has not been calibrated.

 

Then I get the calibration information of IMU data from another ZR300 ROS driver(https://github.com/IntelRealSense/realsense_samples_ros/tree/kinetic-devel/realsense_ros_camera https://github.com/IntelRealSense/realsense_samples_ros/tree/kinetic-devel/realsense_ros_camera), the output of topic /camera/accel/imu_info is like this:

 

data: [1.0236904621124268, 0.0, 0.0, -0.0455622673034668, 0.0, 1.0147671699523926, 0.0, 0.14746201038360596, 0.0, 0.0, 0.9979372620582581, -0.41241949796676636]

 

noise_variances: [0.14709974825382233, 0.14709974825382233, 0.14709974825382233]

 

bias_variances: [9.999999747378752e-05, 9.999999747378752e-05, 9.999999747378752e-05]

I think this 12-dimensions matrix is the calibration info. It can be rewritten like this:

 

data: [ 1.0236904621124268, 0.0, 0.0, -0.0455622673034668,

 

0.0, 1.0147671699523926, 0.0, 0.14746201038360596,

 

0.0, 0.0, 0.9979372620582581, -0.41241949796676636]

/* Scale X cross axis cross axis Bias X */

/* cross axis Scale Y cross axis Bias Y */

/* cross axis cross axis Scale Z Bias Z */

It looks like a Homogeneous transformation matrix. So I use it to calibrate the accel data[accel_x, accel_y,accel_z]'. I'm not sure how to do it. I try a reasonable way:

 

calibrated_accel_x= (accel_x-(-0.0455622673034668))*1.0236904621124268 //(accel_x-Bias X)*Scale X

the other two axes are calibrated in the same way. But it doesn't work. Actually, I have tried all possible combinations to calibrate the raw data. None of them provide a reasonable result.

So is this calibration information of IMU reliable? Is there any SDK to calibrate the raw imu data?

0 Kudos
5 Replies
idata
Employee
415 Views

Hello zed43,

 

 

Thanks for reaching out!

 

 

I would like to point out that there has been another thread about this topic in the community. Why don't you check https://communities.intel.com/thread/116169 https://communities.intel.com/thread/116169 to see the suggestions previously provided in that thread?

 

 

Also, I found this document: https://github.com/ethz-asl/dji_onboard_sdk_ros/wiki/Sensor-Calibration, it explains how to calibrate sensors. On my quick review of the document I read that this guide should performs all necessary calibration procedures such as fish-eye camera intrinsic calibration, fish-eye camera and ZR300 IMU extrinsic calibration, and fish-eye camera and DJI IMU extrinsic calibration. This might also be of help for you.

 

 

I hope this information helps you,

 

Pedro M.
0 Kudos
蔡之奡
Beginner
415 Views

This is not exactly my question. The thing is I have got the IMU intrinsic calibration through SDK, I want to know how to use IMU intrinsic to calibrate IMU raw data. Until now, I prefer the factory calibration to calibration by myself. So my question is how to calibrate IMU raw data by using its intrinsic, or is this intrinsic reliable enough. If this intrinsic isn't right, I have to calibrate by myself.

0 Kudos
idata
Employee
415 Views

I understand, from what I can see in https://github.com/IntelRealSense/librealsense, one of the functionalities of librealsense is to provide intrinsic/extrinsic calibration information. However, I can't find how/if you can use IMU intrinsic to calibrate IMU raw data. Let me see what I can find out about this, if I find anything useful I'll make sure to post it here.

 

 

Pedro M.
0 Kudos
idata
Employee
415 Views

Hi zed43,

 

 

I just want to share the following thread with you, you may find it helpful: https://github.com/IntelRealSense/librealsense/issues/538

 

 

Regards,

 

Diego
0 Kudos
idata
Employee
415 Views

Hello zed43,

 

 

Do you have any updates about this? Did Diego's link helped?

 

 

Pedro M.
0 Kudos
Reply