Enable position estimation without compass

Hey everyone,

Is it possible to enable position estimation without the compass? I know you can set up the EKF_origin using the mission planner, but if you do not have a compass you will not get position. I am flying in an environment where the GPS does not work and the compass may be affected by some steel structures, but I would like to be able to get position from the optical flow sensor. I know that just setting the initial position is not enough since I guess it requires an initial heading. If I am correct, anyone knows how to set it up?

Regards

It only works in the github master branch. Are you using that version?

Thank you @amilcarlucas
I am going to try it today. Currently, I am using the version 4.0.5, and I was about to switch to the version 4.0.6 since these are stable.

Sorry, I missunderstood you were using ArduPlane. On Arduplane operation without compass is possible. On ArduCopter AFAIK it is not yet possible.
And if it is, then with version 4.1.0-dev, not with 4.0.6.

Good to know, theoretically, even if it is not supported, it should be possible as soon as an initial heading is set. Despite the yaw estimation only using the gyroscope is not perfect, it should work with a little drift. And that yaw is in the local earth reference frame from the theory of Euler angles. I wonder if it is something that can be easily changed in the code. I will dig a little bit more in that matter today.

Don not overestimate the noise and drift of the MEM accel and Gyro. These devices are doing fine for stabilisation as the gravity is a fixed and constant vector but for the rest it is just a ball of noise that EKF cannot filter out unless you can fusion magnetometer and other estimators.

As for OpticalFlow, this is a relatively good velocity estimator if you supply good lighting and non reflective and well textured surface. Once again, the position estimation is prone to large error as the integration will add noise and uncertainty from the optical readings and processing algorithm.

You are right @ppoirier , the IMU on the Pixhawk, regardless which one is not the best in the market. May be with one microstrain or something different it may work since the noise of the gyroscopes are less.

Thank you all for your comments!