Does a dual IMU setup make sense for a variable-width rover with laterally offset Pixhawk?

Hello,

I am working on a ground robot approximately 2.20 m wide, with an adjustable track width that can vary by up to 75 cm depending on the configuration.

The system is based on a Pixhawk running PX4, with a companion computer.
The vehicle operates at low speed, up to 6 km/h maximum.

Currently, the IMU is not mounted at the geometric center of the robot but is laterally offset. I am trying to better understand the implications of this configuration.

My questions are:
What is the real impact of a laterally offset IMU on EKF2 state estimation for a ground rover at low speed?
If the track width changes and potentially shifts the center of mass, would it make sense to dynamically update EKF2_IMU_POS_Y via MAVLink when the width changes?
Would installing two IMUs, one on each side of the robot, provide any real estimation benefit, or does PX4 only perform selection/averaging without full geometric compensation?

The goal is to maintain clean and consistent state estimation despite the variable geometry.

Thank you for your insights.

For a rover the IMU position doesn’t really matter, I would not be concerned. And no, don’t install a separate IMU.