I am trying to build a car like robot that uses ackermann steering.

For my robot, I am using DC motors to drive the wheels with wheel encoders attached. For steering, I am using linear actuators to push/pull the wheels in and out (parallel to the ground). I am able to get position feedback from the steering actuators.

I am wondering, in order for me to visualize the baselink in rviz, I can use the wheel encoder data to publish linear translation to the odom tf which will move the baselink back and forth in rviz, but how do I go about publishing yaw data to the base_link?

I know that in a differential drive set up, both linear and angular displacement can be published based on wheel encoder data. Has anyone built a car like robot and successfully visualize the base_link in rviz with accurate representation of vehicle movement from real life?

I would like to avoid the use of IMU if possible, but if not, please let me know.

Thank you, Aaron