MYNT-EYE-S-SDK icon indicating copy to clipboard operation
MYNT-EYE-S-SDK copied to clipboard

IMU calibration

Open pzhu008 opened this issue 5 years ago • 8 comments

Hi, Error occured when I use imu_utils to calibrate IMU. (no results for gyro)

I recorded a rosbag of IMU raw data for 90 mins. All gyro_x, gyro_y, gyro_z are zeros. Then I use ./record to generate a dataset. Also all gyro_x, gyro_y, gyro_z are zeros.

Do you have any idea about this?

Thanks,

imj

pzhu008 avatar Nov 08 '19 23:11 pzhu008

It seems that there's is some filtering being done setting the gyro values to zero when there is little movement. This makes it impossible to measure the bias and bias random walk.

It would be more appropriate to send the raw data as the topic says.

I think the firmware had to be adjusted for this, could you please make this happen?

I'm experiencing some issues in maplab where the frame gets ignored because the gyro data is zero, also the gyro intrinsics can't be measured for callibration purposes.

niekez avatar Jan 17 '20 09:01 niekez

It seems that there's is some filtering being done setting the gyro values to zero when there is little movement. This makes it impossible to measure the bias and bias random walk.

It would be more appropriate to send the raw data as the topic says.

I think the firmware had to be adjusted for this, could you please make this happen?

I'm experiencing some issues in maplab where the frame gets ignored because the gyro data is zero, also the gyro intrinsics can't be measured for callibration purposes.

pzhu008 avatar Jan 17 '20 17:01 pzhu008

try SDK 2.3.6 without filter

pzhu008 avatar Jan 17 '20 17:01 pzhu008

try SDK 2.3.6 without filter

It's not possible to downgrade the firmware for 2.3.6. Also I'm using the S1030, and as far as I know there's no processing being done by the SDK. I think it's happening on a firmware level, but maybe someone can point me to the SDK code where filtering is done on the gyro.

niekez avatar Jan 20 '20 11:01 niekez

Hi, any update on this issue? I'm using the example noise values given in the repository here but I would very much prefer calibrating my mynteye myself...

lesurp avatar Aug 04 '20 09:08 lesurp

Actually, this filtering is actively deteriorating the processing I am doing with the values. Is there any reason why the filters cannot be disabled on the s1030, while it can on the other devices? (https://slightech.github.io/MYNT-EYE-S-SDK-Guide/src/control/imu_low_pass_filter.html)

lesurp avatar Aug 06 '20 09:08 lesurp

Sorry,the company is dead..

| | 熊若阳 | | 邮箱:[email protected] |

签名由 网易邮箱大师 定制

On 08/06/2020 17:12, Paul Lesur wrote:

Actually, this filtering is actively deteriorating the processing I am doing with the values. Is there any reason why the filters cannot be disabled on the s1030, while it can on the other devices? (https://slightech.github.io/MYNT-EYE-S-SDK-Guide/src/control/imu_low_pass_filter.html)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

KalmanXiong avatar Aug 06 '20 09:08 KalmanXiong

This is a bug in SDK version 2.5, you can ref to #113

LSXiang avatar Jan 28 '21 09:01 LSXiang