Fusing output of imu_filter_madgwick with vehicle odometry in robot_localization package
Hi everyone. I am using imu_filter_madgwick to get an orientation estimate from my IMU data. The covariance matrix that imu_filter_madgwick supplies basically has all zero entries. Now, I want to fuse this with the odometry topic from my robot (which also has zero entries as its associated covariance values) in the robot_localization package. I plan to use only the x and y velocities from the vehicle odom topic as the x and y estimates are derived by integrating the velocities and there is no point feeding the same information twice.
1.What impact does providing topics with zero covariances will have on the filtered odometry? 2. In theory, is the information that I am providing i.e. orientation from the imu_filter_madgwick and linear velocities from the odom topic (by setting corresponding values to true in the config matrix in robot_localization) enough to obtain the filtered odometry ?
