Zero out IMU and rotor fitness factors

Dear Arducopter society,

This question is about quadrotors or any flying machine but the flight controls is not Arducopter. I have built my own system using raspberry pi and MCU.

Everything is checked multiple times like propeller directions, motor-propeller combinations IMU’s etc.

The problem is all four motors are not identical. I have used 4 motors from same vendor. I did a quite good estimation/identification for thrust by using RC testbench. However, still due to motor different properties there is a drift in UAV.

One thing is to do some observations on each motor while unstable hover and premultiply motors by some relative percentage values. I tried this it reduced the drift but still uav is drifting.

Is there some algorithm on this?
How Ardupilot solves this problem?

The other things IMU I am using also has a resolution of 1 degree and also testing my data if put static the values have some 0.5 degree of offset, not completely zero. This also a potential cause of drift.

Is there any way to correct this IMU error either by using Vicon motion capture system or others.

ArduCopter uses sensor fusion in an EKF to solve these issues.

1 Like

Could you please send me some documentation that how EKF is used and what sensor measurements are there. I used kalman filter to get IMU quaternion orientations which are very stable.

Now please tell me any source to move forward?
I have vicon, IMUs available.

This one is a good starting point: EKF — Dev documentation

Also try to study it from source code…

1 Like

Thank you. I will give it a try