Why is just one accelerometer used?

I wonder why the values of >1 accelerometers do not get averaged, and instead just one is used?

Do you want to explain your question better? I’m not sure what you are talking about.

I often wondered why there isn’t an averaging mode in this and other cases of multiple sensors of one quantity. It needs someone to try it out :wink:
You would probably need runtime weights to give to each sensor. Selecting the weights is one tricky bit and the way it is done probably would need to be varied dependent on the quantity you are measuring

Well, I guess I might be wrong. In the EKF is looks as if a new instance is generated for each IMU.

In the EKF2 we have an instance per IMU, changing between them when needed. EKF1 has only one core and attributes weights for two IMUs. I’ll defer to others to answer why an instance per IMU is better.

I believe you can find answers here: Enable separate EKF2 instances for each IMU by priseborough · Pull Request #3125 · ArduPilot/ardupilot · GitHub

Thanks, I was just interested why and how it works exactly.