Covariance is growing unbounded in robot_localization

edit

Hi, I'm building a custom robot (3 wheel, differential drive). It has an IMU, reporting acceleration, angular velocity and magnetic field; wheels encoders that reports wheels velocities; a GPS reporting position. I'm running two instance of RL, the first one for odm->base_link use imu and wheel encoders:

 odom0: /chori/odom
  odom0_config: [false, false, false,
                 false, false, false,
                 true,  true,  false,  #true,  true,  true,
                 false, false, false,
                 false, false, false]
  odom0_queue_size: 10
  odom0_nodelay: true
  odom0_differential: false
  odom0_relative: false

# imu configure:
  imu0: /imu/data
  imu0_config: [false, false, false,    # x, y, z,
                false, false,  true,    # roll, pitch, yaw
                false, false, false,    # vx, vy, vz
                false, false,  true,   # vroll, vpitch, vyaw
                true,  false,  false]    # ax, ay, az  
  imu0_nodelay: false
  imu0_differential: false
  imu0_relative: false
  imu0_queue_size: 10

I just trust on encoders velocities, the yaw (imu_filter_madwick), the v_yaw (giroscope) and linear acceleration in X axis.

Then I have the second instance tha use the same sensors but adds the gps published by the navsta_transform node.

I have two main odometry messages, /odometry/filtered and /odometry/filtered_map . The second one reporting map->base_link odometry. I was expecting that map->base_link message have a bounded value in their covariance because of the GPS. But it's not. Position covariance slowly grow unbounded. I'm wondering why is that?

thank and best regards