"Ball Magnet" Roller System I am trying to design a rolling 'ball magnet' system similar to this video as a prototype for a specialized version of it I want to try on a robot this year: I want to gain a solid understanding of all the factors for constructing something like this and the underlying principles. The ball I want to use for this can deform reasonably; it has a foam core, but it doesn't compress much and is covered in a ribbed plastic, which should have fairly good grip. The floor this would be on is carpets, with I'm thinking likely has a pretty high friction. Those guys: http://www.macfloor.com/carpet/brands/shaw/Neyland-II/Neyland-II.html How would i calculate the speed which my roller needs to spin the ball? The part I am having difficulty with seeing how the coefficient of friction of the ball and roller factors in, accept for changing the maximum speed at which you can roll the roller without exceeding static friction. I have control over the material I use on the roller, so what is a desirable μ and material? After I understand how to make a roller like this I want to create my actual device, a roller which directionally slides the ball across its length, and pushes it out in front of the robot. I was thinking of something almost cone shaped, the narrower head being the direction the ball was to travel. This design is based on the concept that when the rollers and the ball contact, the ball is pulled toward the rollers, and so it will move to the narrower end since it has a force pulling it closer to the robot, and the tip of the cone is closer to the robot. Is this premise correct, and how should I orient this design to get the most forward momentum and the least.