There have been a lot of great discussion threads regarding how to design/build the best shifting transmission. These discussions almost imply that it is a foregone conclusion that shifting is better. I personally do not agree with this point of view and think the topic merits discussion.
Maybe you don’t need to shift at all.
I would argue that for the last two years shifting was not necessary. If you can attain your desired top speed and still have enough low end torque to break loose all of your tires/tracks in a pushing match you do not need to shift. Electric motors unlike internal combustion engines make their max torque at stall, which is where you want the most torque and pushing force. For max speed you need enough torque to go only as fast as you can drive. For us this is about 10 ft/sec. At neither extreme do you need to be at the max power output of your motors.
In 2001 and 2002 we could pick up movable goals. In 2002 a robot could pick up part or all of the two 90# goals at once and drag them back to their scoring zone while dragging another robot backwards. This was a season to have a shifting robot. For the past two years we did not have the ability to transfer any additional weight onto our robots during a match. We will make the shifting /non-shifting decision after the game is announced.
What are the risks rewards of having a shifting transmission?
Risks: Reliability, robustness, weight, packaging space, cost, complexity, time and energy required, etc.
Rewards: Maximum low end torque / top end speed.
A transmission is sexy and lots of fun to design and build. We have built three different two speed transmissions and a CVT. We have never put a shifting transmission on the field. In the last two years we have played over 200 matches (in season and post season) and have never lost a match because of a transmission problem.
Many teams do not really have the resources to put a shifting transmission on their robot but will proceed to attempt it anyway. This can divert scarce human and financial resources away from the other robot systems that need them: ball grabbers, hanging devices, etc. In many cases, I feel that this should not be done.