Used the robot pose taken from the costmap in the odom frame (TF) to get the current position of the robot (instead of the odom topic.
The rest of Twist and covariance values of the odom topic are still used.