We are thinking about using Odometry to improve our autonomous and I got the following two questions, would really appreciate some help.
Looked like many teams would install the two parallel encoders on separate wheels instead of the driving wheel, they also added rubber band to ensure the additional wheels alway touch the ground. Do we know why? It seemed to us that directly install the encoder on driving wheel would be a simpler solution.
It’s recommended to use 3 wheels for Odometry, has anyone tried two wheel Odometry and how does it work?
You band the tracking wheels into the ground because if the robot leans forward or backward, drive wheels may come out of contact with the ground, but the banded tracking wheels will not.
For our last robot we used one wheel odometry: We had a gyro to measure rotation, and a forward-backward tracking wheel. We had tractions on our drive, so no need for a second wheel.
In addition, putting the shaft encoders on the driven wheels makes them useless, if you want to do that, using the IME (integrated motor encoders), that a lot of people find sufficient for PID. The idea with putting them separately with 2.75 inch omnis is to negate any slop caused by attaching it to a driven wheel/motor. Also both two wheel (perpendicular and parallel + inertial sensor (out of stock)) and three wheel (two parallel and one perpendicular) work just fine. The two wheel works because instead of having a second parallel wheel to calculate heading, you use the inertial sensor.