[KenWittlief]

from my experience, when you try to damp or smooth the controls from the operator to the motors, all you end up doing is making the robot response sluggish and spongy

if the robot is twitchy and jerky then it requires a smooth and steady hand on the controls.

The motion of the robot is not linear. if you map the joystick inputs to the actual motion of the robot, you will discover the static and kinetic friction effects, when you start from a stand still it takes considerable input to get the bot to move, then once it starts moving it takes less input to hold a steady or low speed

but if you damp or average the driver inputs, then when the bot starts to move it will take some time before the driver can get the motors to back off.

The control of the robot is a closed loop feedback system. For the simplist robots the feedback is the drivers eyes, and the control is his brain: the driver moves the joystick, sees what the robot does (visually) and then makes adjustments

damping the response only makes the robot sluggish. If your driver has practice and is good, then he can make a fast, twitchy, responsive robot dance across the floor precisely

the answer to your question is to put sensors on the robots, and make the feedback control electro-mechanical, so that the robot itself can 'see' how fast its moving (through sensors) and compair that to what the operator is asking for, and make the corrections itself.

The reason this works better is the controller on the robot can respond in about 1/30th of a second to sensor feedback. The best a human driver can do, using eye to hand coordination, is about 1/10th of a second. And thats his best. The controller will response in 1/30th of a second to every sensor or driver input, consistanly and accurately.

Closed loop feedback. PID control loop. Thats the best way to control a FIRST robot.