kriswiner
commented
7 years ago It is really quite simple. The Madgwick filter requires the data in one of the standard conventions for AHRS. I usually use NED (North, East, Down) but ENU will also work. The proceudre is to choose which axis you want to poit int he North direction. Let's say Ax (x-axis acceleration). Then depending on the sensor axes orientation of your system, you can choose which axes are pointed in the North direction, which the East, and which the Down. For the MPU9250 with Ax pointing in the nNorth direction, then Gx is also North, My is North, so the madwich filter should receive data as:
M(Ax,..,.., Gx,..,.., My,..,..);
East is -Ay, -Gy and -Mx. Down is -Az, -Gz, and Mz.
So the full function ought to be:
M(Ax, -Ay, -Az, Gx, -Gy, -Gz, My, - Mx, Mz);
I usually negate the A components since gravity == +1 is up by convention.
This process works for any sensor combination.
On Fri, Oct 6, 2017 at 2:37 AM, lyravega notifications@github.com wrote:
Hello Kris, thank you so much for the 9250 stuff. I've been trying to build a DIY head-tracker, and I wouldn't even tried it without your code š
I have a question though. I've seen a few example DIY stuff based on your code, and in all of those, including your MPU9250 only code, Madgwick is used as below:
MadgwickQuaternionUpdate(ax, ay, az, gxPI/180.0f, gyPI/180.0f, gz*PI/180.0f, my, mx, mz);
However, passing the values like this causes heading (yaw) to be heavily affected by the attitude&bank (pitch&roll). While the device is on flat surface, heading is correct but tilt it slightly and it goes nuts.
So I've been reading on issues for all of your code (pretty much anything 9250 related that you've shared with the world, so to speak), and in one of them you've replied to a similar issue with this:
MadgwickQuaternionUpdate(-ay, -ax, az, gyPI/180.0f, gxPI/180.0f, -gz*PI/180.0f, mx, my, mz);
This one, while provides a stable heading when the "compass" is tilted, swaps bank with attitude. I can swap it again, but haven't been able to test it that much as I've tried another, from "MPU9250_MS5637_AHRS_t3.ino", which passes this:
MadgwickQuaternionUpdate(-ax, ay, az, gxPI/180.0f, -gyPI/180.0f, -gz*PI/180.0f, my, -mx, mz);
Now this one is very good, seems to be the best one for my case, quaternion update works as expected and tilting the compass doesn't affect the heading that much; it's still affected (about 1-2 degrees), but I blame myself for it.
So my question is, which one should I use? The one that gives me the accurate readings, of course... However I am very confused right now as different 9250s seem to have different parameters to pass.
Oh almost forgot. The last 3 parameters, why are the passed ones are y-x-z and not x-y-z? What I'm trying to ask is, the function is void MadgwickQuaternionUpdate(float ax, float ay, float az, float gx, float gy, float gz, float mx, float my, float mz), but we pass ..., my, -mx, mz); is it because of some quaternion related stuff? Apologizes for stupid questions/inexperience/etc. Keep up the great work!
ā You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/kriswiner/MPU9250/issues/192, or mute the thread https://github.com/notifications/unsubscribe-auth/AGY1qrZ_U8fy6U984lJq7g4DtYEEnAd1ks5spfTAgaJpZM4PwO01 .
lyravega
Hello Kris, thank you so much for the 9250 stuff. I've been trying to build a DIY head-tracker, and I wouldn't even tried it without your code :+1:
I have a question though. I've seen a few example DIY stuff based on your code, and in all of those, including your MPU9250 only code, Madgwick is used as below:
MadgwickQuaternionUpdate(ax, ay, az, gx*PI/180.0f, gy*PI/180.0f, gz*PI/180.0f, my, mx, mz);However, passing the values like this causes heading (yaw) to be heavily affected by the attitude&bank (pitch&roll). While the device is on flat surface, heading is correct but tilt it slightly and it goes nuts.
So I've been reading on issues for all of your code (pretty much anything 9250 related that you've shared with the world, so to speak), and in one of them you've replied to a similar issue with this:
MadgwickQuaternionUpdate(-ay, -ax, az, gy*PI/180.0f, gx*PI/180.0f, -gz*PI/180.0f, mx, my, mz);This one, while provides a stable heading when the "compass" is tilted, swaps bank with attitude. I can swap it again, but haven't been able to test it that much as I've tried another, from "MPU9250_MS5637_AHRS_t3.ino", which passes this:
MadgwickQuaternionUpdate(-ax, ay, az, gx*PI/180.0f, -gy*PI/180.0f, -gz*PI/180.0f, my, -mx, mz);Now this one is very good, seems to be the best one for my case, quaternion update works as expected and tilting the compass doesn't affect the heading that much; it's still affected (about 1-2 degrees), but I blame myself for it.
So my question is, which one should I use? The one that gives me the accurate readings, of course... However I am very confused right now as different 9250s seem to have different parameters to pass.
Oh almost forgot. The last 3 parameters, why are the passed ones are y-x-z and not x-y-z? What I'm trying to ask is, the function is
void MadgwickQuaternionUpdate(float ax, float ay, float az, float gx, float gy, float gz, float **mx**, float **my**, float **mz**), but we pass..., **my**, -**mx**, **mz**);is it because of some quaternion related stuff? Apologizes for stupid questions/inexperience/etc. Keep up the great work!