search

kriswiner / MPU9250

Arduino sketches for MPU9250 9DoF with AHRS sensor fusion
1.03k stars 471 forks source link

When the IMU rotates about one axis, angles about other two axes change too? #26

Open britam opened 8 years ago

britam commented 8 years ago

Hi, i modify and run the code of the sketch of kriswiner, and i have the problem what when i rotate the IMU (i use the MPU9250 and Arduino Pro mini 3.3 volts 8 Mhz) about one axis angles obout the other two change too. HOW CAN I SOLVE THAT????? and why happend that?

thaks!!!!

kriswiner commented 8 years ago

This can happen if your fusion rate is small compared to your sample data rate (likely with the 8 MHz Pro Mini) and/or your sensors are not properly bias calibrated. See the MPU6050 Wiki for more info about these.

britam commented 8 years ago

i'm gonna change my bad redacted question, indeed the angles seems good calculated, i was thinking what each angles are represented independent of the other two angles. the general function i see in the program is what when i move my protoboard in a horizontal plane, the yaw angle change correctly, but the other two angles (pitch and roll) dont change (that is what i want), but when i rotate in the other planes where i only wish change only pitch, or only roll .... all the angles change. so, my question is: that is the real behavior of the sketch ..... Thanks for your fast answer and for your code :)

kriswiner commented 8 years ago

If you rotate about only one axis at a time, you should see only changes in that one parameter, i.e., pitch or roll. If you have a compound angle then rotations will appear to change all the values. In other words, if you point up at 45 degrees and then only rotate about the pitch angle, you will see heading change too. This is expected and a result of the Euler angle approximation to orientation.

-----Original Message----- From: britam [mailto:notifications@github.com] Sent: October 9, 2015 9:26 AM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

i'm gonna change my bad redacted question, indeed the angles seems good calculated, i was thinking what each angles are represented independent of the other two angles. the general function i see in the program is what when i move my protoboard in a horizontal plane, the yaw angle change correctly, but the other two angles (pitch and roll) dont change (that is what i want), but when i rotate in the other planes where i only wish change only pitch, or only roll .... all the angles change. so, my question is: that is the real behavior of the sketch ..... Thanks for your fast answer and for your code :)

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-146919701 . https://github.com/notifications/beacon/AGY1qlcs79RHarST7AdaRVlltkjTCx52ks5 o5-I0gaJpZM4GLv6x.gif

britam commented 8 years ago

Ok, i understand now..... but i need what the three angles moves in a independent way, maybe the transformation is not good enough for what i looking for. I know what quaternions angles are the absolute format, how can i transform them to a format what belongs to me calculate the movement of a body in the space with three angles what move in an independent way like the navigation angles or Tait–Bryan angles

kriswiner commented 8 years ago

Euler angles are Tait-Bryan angles. The problem is these angles are appropriate for spherical surfaces, not Euclidean space.

-----Original Message----- From: britam [mailto:notifications@github.com] Sent: October 9, 2015 10:59 AM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

Ok, i understand now..... but i need what the three angles moves in a independent way, maybe the transformation is not good enough for what i looking for. I know what quaternions angles are the absolute format, how can i transform them to a format what belongs to me calculate the movement of a body in the space with three angles what move in an independent way like the navigation angles or Tait-Bryan angles

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-146948292 . https://github.com/notifications/beacon/AGY1qskNxW5PsCFNdQ1yk5P0_cuP47Q5ks5 o5_fngaJpZM4GLv6x.gif

britam commented 8 years ago

understood, thanks!!!! ;) if i make an advance with my proyect i'll post it here.

britam commented 8 years ago

Hi Kris, Finally, i now can read the three angles properly. But my last problem is what when i use the madgwick algorithm the angles are updated slowly. What parameters i must change to make the filter works properly, or what filter you recomend to me use the mahoney or the madgwick. Regards!!!!

kriswiner commented 8 years ago

What did you have to do to get the angles to behave? Some others report this behavior and I see it myself often, did you change the code somehow?

You may use either Madgwick or Mahony. The best way to get faster update speeds is to increase the speed of your controller. Are you using a Teensy? If so you can run it at 96 MHz.

Kris

-----Original Message----- From: britam [mailto:notifications@github.com] Sent: October 29, 2015 2:53 PM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

Hi Kris, Finally, i now can read the three angles properly. But my last problem is what when i use the madgwick algorithm the angles are updated slowly. What parameters i must change to make the filter works properly, or what filter you recomend to me use the mahoney or the madgwick. Regards!!!!

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-152338922 . https://github.com/notifications/beacon/AGY1qsP-7dCa6IhPnXzYmjMve5Q_sWxmks5 pAoy6gaJpZM4GLv6x.gif

yerzhik commented 8 years ago

Having the same issue. When sensor is rotated along Z axis while being prependicular to the floor (Z axis is prependicular to floor) then its perfectly rotating and angles showing correct values. But when I start rotating along x or y axis then other two axis starts rotting too. So the question is were you able to resolve this problem? What is the reason?

kriswiner commented 8 years ago

I always have the same question when people ask me about this: how are you calibrating your sensors? In particular, how are you calibrating your magnetometer?

-----Original Message----- From: yerzhik [mailto:notifications@github.com] Sent: January 20, 2016 8:32 PM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

Having the same issue. When sensor is rotated along Z axis while being prependicular to the floor (Z axis is prependicular to floor) then its perfectly rotating and angles showing correct values. But when I start rotating along x or y axis then other two axis starts rotting too. So the question is were you able to resolve this problem? What is the reason?

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-173448569 . https://github.com/notifications/beacon/AGY1qhWU6CQdUoKrtBtzCjaRM6p8ELXPks5 pcFa7gaJpZM4GLv6x.gif

yerzhik commented 8 years ago

For acceleration I estimate the bias and set it. There is a DMP code for that (motion application library from invensense) And I set up the DMP to calibrate the GYRO. (actually inside the mpu init function) Which does it (as it's written) for 8 seconds and the MPU should be set so that pins are up or down (initial position). Does that makes sense? As for magnetometer I'm afraid it was not calibrated (at least if only mpu init function doesn't do it somehow)

kriswiner commented 8 years ago

How are you calibrating the mag? If it is uncalibrated you will just get nonsense results.

-----Original Message----- From: yerzhik [mailto:notifications@github.com] Sent: January 20, 2016 11:35 PM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

For acceleration I estimate the bias and set it. There is a DMP code for that (motion application library from invensense) And I set up the DMP to calibrate the GYRO. Which does it (as it's written) for 8 seconds and the MPU should be set so that pins are up or down (initial position). Does that makes sense?

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-173483744 . https://github.com/notifications/beacon/AGY1qjwp95IQNn9AjeuNmL4RwAAesF0Dks5 pcIGOgaJpZM4GLv6x.gif

yerzhik commented 8 years ago

I'm not, but from Invensense documentation on offset registers I see only Accel/Gyro offset registers (for setting up bias). But how would I tell DMP to use offsets for magnetometer?

kriswiner commented 8 years ago

I don't know, but this is one of many reasons I do not use the DMP.

-----Original Message----- From: yerzhik [mailto:notifications@github.com] Sent: January 21, 2016 10:27 PM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

I'm not, but from Invensense documentation on offset registers I see only Accel/Gyro offset registers (for setting up bias). But how would I tell DMP to use offsets for magnetometer?

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-173823904 . https://github.com/notifications/beacon/AGY1qjdh9mOpQyktQj4uSgbZfdL073NJks5 pccM0gaJpZM4GLv6x.gif

yerzhik commented 8 years ago

I found out that the issue was in that the gyroscope was still calibrating while I started rotating it. Waiting for 15 seconds makes much better results. But issue now happens only if I rotate only along one axis until 90 degree. Is it related with euler angles or it shouldn't be like that at all? Getting euler angles from quaternions and didn't expect that to happen.

On Fri, Jan 22, 2016 at 11:03 PM, Kris Winer notifications@github.com wrote:

I don't know, but this is one of many reasons I do not use the DMP.

-----Original Message----- From: yerzhik [mailto:notifications@github.com] Sent: January 21, 2016 10:27 PM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

I'm not, but from Invensense documentation on offset registers I see only Accel/Gyro offset registers (for setting up bias). But how would I tell DMP to use offsets for magnetometer?

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-173823904 . < https://github.com/notifications/beacon/AGY1qjdh9mOpQyktQj4uSgbZfdL073NJks5 pccM0gaJpZM4GLv6x.gif>

— Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-173978915.

kriswiner commented 8 years ago

You should expect some cross talk between yaw , pitch and roll. It's hard to get away from unless you specifically correct for it.

-----Original Message----- From: yerzhik [mailto:notifications@github.com] Sent: January 25, 2016 9:28 PM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

I found out that the issue was in that the gyroscope was still calibrating while I started rotating it. Waiting for 15 seconds makes much better results. But issue now happens only if I rotate only along one axis until 90 degree. Is it related with euler angles or it shouldn't be like that at all? Getting euler angles from quaternions and didn't expect that to happen.

On Fri, Jan 22, 2016 at 11:03 PM, Kris Winer notifications@github.com wrote:

I don't know, but this is one of many reasons I do not use the DMP.

-----Original Message----- From: yerzhik [mailto:notifications@github.com] Sent: January 21, 2016 10:27 PM To: kriswiner/MPU-9250 Cc: Kris Winer Subject: Re: [MPU-9250] When the IMU rotates about one axis, angles about other two axes change too? (#26)

I'm not, but from Invensense documentation on offset registers I see only Accel/Gyro offset registers (for setting up bias). But how would I tell DMP to use offsets for magnetometer?

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-173823904 . <

https://github.com/notifications/beacon/AGY1qjdh9mOpQyktQj4uSgbZfdL073NJks5 pccM0gaJpZM4GLv6x.gif>

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-173978915.

Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/26#issuecomment-174839621 . https://github.com/notifications/beacon/AGY1qgNP4uK_ZqQwxPltdCGAHkczpl2Sks5 pdvt1gaJpZM4GLv6x.gif

yerzhik commented 8 years ago

Thank you, I think its solved for me now

JuanPabloGomez commented 7 years ago

I would like to answer Kris Winer's question because I also experienced changes in yaw with changes in the pitch/roll.

What did you have to do to get the angles to behave? Some others report this behavior and I see it myself often, did you change the code somehow?<

I believe the problem happens when sending the mx, my and mz raw values to the Madgwick/Mahoney quaternion function. In my MPU9150, the accelerometer/gyro and magnetometer have difference axis. X in the accel corresponds to Y in the magnetometer. Y in the accel corresponds to X in the magnetometer. Finally, Z corresponds to -Z. This last one is important because the code I have seen accounts for the changes in X and Y but not for the change in direction in Z, as observed below:

MadgwickQuaternionUpdate(ax, ay, az, gx * PI / 180.0f, gy * PI / 180.0f, gz * PI / 180.0f, my, mx, mz);

As you can see, the my and mx are correctly swapped, but there is no change in the sign of mz. The correct code would look like this:

MadgwickQuaternionUpdate(ax, ay, az, gx * PI / 180.0f, gy * PI / 180.0f, gz * PI / 180.0f, my, mx, -mz);**

This solved the problem for me. Hope it helps others.

PatelKishanJ commented 6 years ago

Hello, I need to know about declination-angle,which is useful in measuring YAW. Any formula to find that angle for particular location?

Thanks in advanced.

kriswiner commented 6 years ago

This is measured not calculated. Look it up for your location.

On Tue, Jan 9, 2018 at 2:19 AM, jenextech notifications@github.com wrote:

Hello, I need to know about declination-angle,which is useful in measuring YAW. Any formula to find that angle for particular location?

Thanks in advanced.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/kriswiner/MPU9250/issues/26#issuecomment-356241955, or mute the thread https://github.com/notifications/unsubscribe-auth/AGY1qsBw6YAZW1Ejgt-W1FiXZ_Q5DaM7ks5tIz1HgaJpZM4GLv6x .

PatelKishanJ commented 6 years ago

Okay. I have checked it for my location. It looks like as i have attached picture.

image

 This is in different format as we put declination angle's value in code.

Can you tell me that from above image what should declination angle instead of we have used in code (yaw -= 23.45f; // Declination at Danville, California is 13 degrees 48 minutes and 47 seconds on 2014-04-04)

Please,confirm this one.

kriswiner commented 6 years ago

0.49 degrees?

On Thu, Jan 11, 2018 at 1:59 AM, jenextech notifications@github.com wrote:

Okay. I have checked it for my location. It looks like as i have attached picture.

[image: image] https://user-images.githubusercontent.com/26138454/34819499-e89787e8-f6e3-11e7-8993-40d2074e2214.png

This is in different formate as we put declination angle's value in code.

Can you tell me that from above image what should declination angle instead of we have used in code (yaw -= 23.45f; // Declination at Danville, California is 13 degrees 48 minutes and 47 seconds on 2014-04-04)

Please,confirm this one.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/kriswiner/MPU9250/issues/26#issuecomment-356884951, or mute the thread https://github.com/notifications/unsubscribe-auth/AGY1qjZ4FdjoGVWn-aGusLw4-IqxlqIBks5tJduJgaJpZM4GLv6x .

PatelKishanJ commented 6 years ago

Hello Kris, When i tried to get declination angle,it shows as like image.

So,what should i put value of it instead of 23.45.

kriswiner commented 6 years ago

The declination is a propery of your location on Earth, not the sensor. For your area, 0.49 degrees....

On Thu, Jan 11, 2018 at 8:46 PM, jenextech notifications@github.com wrote:

Hell Kris, When i tried to get declination angle,it shows as like image.

So,what should i put value of it instead of 23.45.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/kriswiner/MPU9250/issues/26#issuecomment-357142276, or mute the thread https://github.com/notifications/unsubscribe-auth/AGY1qrIQnvzgXSR3p3JkUsuIJ70oRsyCks5tJuOugaJpZM4GLv6x .

PatelKishanJ commented 6 years ago

Okay. Thanks for fast response. So,now i will check all data by considering declination angle as 0.49. Right?

liutianxin0722 commented 3 years ago

Hi Kris, Finally, i now can read the three angles properly. But my last problem is what when i use the madgwick algorithm the angles are updated slowly. What parameters i must change to make the filter works properly, or what filter you recomend to me use the mahoney or the madgwick. Regards!!!!

Hello britam！ I meet the same problem that when I rotate about X axis，the values in Y axis also change. I use Mahoney algorithm to obtain eular angles from a 6-axis MPU 6050 sensor. How did you solve the problem? Thank you

kriswiner commented 3 years ago

You need to send the sensor data to the filters as NED, so aN, aE, aD, gN, gE, gD to get consistent results.

On Tue, Sep 22, 2020 at 3:55 AM liutianxin0722 notifications@github.com wrote:

Hi Kris, Finally, i now can read the three angles properly. But my last problem is what when i use the madgwick algorithm the angles are updated slowly. What parameters i must change to make the filter works properly, or what filter you recomend to me use the mahoney or the madgwick. Regards!!!!

Hello britam！ I meet the same problem that when I rotate about X axis，the values in Y axis also change. I use Mahoney algorithm to obtain eular angles from a 6-axis MPU 6050 sensor. How did you solve the problem? Thank you

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/kriswiner/MPU9250/issues/26#issuecomment-696647806, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABTDLKSQZA3J6UNUDXWUYCDSHB7BBANCNFSM4BRO72YQ .