I like the idea of using one or more belts to turn the center wheel, and I drew it up similar to what @wgorgen suggested (correct me if I’m wrong) where the wheel is driven from the motor on the right side of the picture and is actuated down using the same pneumatic mechanism that the original used, just angled about 5 degrees to the right.
I am thinking about doing something like @GeeTwo suggested and moving the center wheel closer to one side but I’m worried that doing so will move the robot’s center of rotation when the wheel is down and mess up its position tracking in auto. Will this be the case?
One of the objectives of this project is to maximize pushing power when the center wheel is engaged, and since mecanums only transmit ~70% of torque forward because of the angled rollers, it makes the most sense to me to take as much weight as possible off of them and onto the traction wheel. Using the pneumatic cylinders in the drawing (1 on each side) gives a total force against the ground of 94 lbs. Assuming the robot is 150 lbs, this means that the total efficiency would be ((94x1)+(0.70x56))/150 = ~89% efficiency. If the wheels were moved closer to one end, the amount of force they could exert would decrease when they would eventually be able to lift up that side of the robot (the cylinder has enough throw for the center wheel to extend .125" below the plane of the two mecanums). The point at which this occurs would naturally depend on the robot’s CoM. Is my reasoning and math correct or does it not work like this?
Finally, in regards to using an omni wheel in the center, I have always had the notion (correct or not) that omni wheels are worse at pushing straight forward than normal traction wheels. Is this true?