If you're trying to estimate all 6 parameters as states, I think you'll find you need more that the IMU because there is no way to estimate the biases of these sensors with just their measurements. You need some other measurements to differentiate the biases from the actual sensing.

Hi,

i have it. :) but i suggest the Quaternion based sensor fusion for IMU. :)

i attach 3 file that is consist of:

1-DCM : for learn about how can u use direction cosine matrix for get orientation

2-Madgwick filter report

3-madgwick filter source code

4-kalman filter source code

for kalman filter, you just need crate 3 independent children from proposed class inc++.

i used this class in ARM micro controller to get roll,pitch and yaw estimation. you just need accelerometer and gyroscope data. just remember that gyro data should be in deg/s. :)

I've written a short document - and accompanying code - on how to perform various types of state estimation (including Kalman filtering) for a simple 6-DOF IMU, such as the MPU-6050.

It is available here: http://philsal.co.uk/projects/imu-attitude-estimation

I hope that helps!

Philip

Hello Philip Salmony is there even a paper about a quaternion-based ekf or ukf filtered algorithm for an imu (gyro&acc only) ? Because all papers I find usually use at least magnetometers or optical/camera elements in the measurement model...

look

https://www.youtube.com/watch?time_continue=2&v=FwQG_DPOCVY

Waldri Oliveira my brasilian is almost non-existent :D but I tried to download from your homepage and I only get error 404 can you please link me the documents? sorry :(

Edit: I found maybe some documents through your profile but it seems you didnt use an extended kalman filter oder unscented. The usual direct angle calculation and kalman filtering this is explained by Philip Salmony. I need quaternion based EKF or UKF algorithms :)

Dear Oleksandr Müller,

since both gyroscope and accelerometer sense dynamic motion in their own local frame, its integration without another sensor which refers to the "absolute" or "reference" frame might not be of great help.

Nonetheless, I suggest that you can refer to Article Quaternion kinematics for the error-state KF

Best,

Daniel

good afternoon, there is some module with inertial sensors and kalman filter included?

Hi again,

I've written a new document and code for attitude estimation using an Extended Kalman Filter (EKF) using a quaternion-based attitude representation, see:

http://philsal.co.uk/quaternion-based-extended-kalman-filter-for-fixed-wing-uav-attitude-estimation

However, this uses more sensors than just gyroscopes and accelerometers - but it can be adapted fairly easily to use less sensors, should not all be available. Let me know if I can be of any help.

Best,

Philip

Hi. You can check out a very detailed 1D and 2D version of the Kalman filter, it helped me a lot to understand and implement one http://www.ilectureonline.com/lectures/subject/SPECIAL%20TOPICS/26/190

P.S.: I am well aware that the post is 8 years old, but it might be still helpful for those who just found that post :)