So i am a beginner in the field of robotics, and I have successfully created an autonomous two-wheeled robot equipped with two encoders. I've implemented a PI regulation on speed, and while the robot performs reasonably well, I still encounter some measurement errors. To enhance its performance, I'm considering the implementation of a Kalman filter.
While I have a theoretical understanding of the Kalman filter, I'm a bit confused about the practical implementation. Should I apply a simple Kalman filter to each encoder independently to smooth the measurements, or is it more appropriate to use the filter on the x and y positions of the robot? . Any guidance on the best approach for implementing the Kalman filter in this context would be greatly appreciated.