[GeeTwo]

I'm not sure if you'd consider the system we're using an "active" intake or not. There are no rollers or such "actively" pulling the pieces to the robot. We considered such rollers, but we decided to tackle clearing the landfill as a primary goal, hoping that we could get quicker at that than the HP can feed totes. Since the landfill is so tightly packed, most active systems would be unable to get around the tote well enough to get an initial grip.

As an alternative, we are implementing an extensive (at least for us) array of sensors so that all the driver has to do is slew into a rough position and let automation take over. We'll have a pair of rangefinders so we can square up on a tote, do some vision processing to find the distinctive tote features and align left to right, and have some touch sensors on our long-side grappler (similar to the Greenhorn's Ri3d, but with a passive "foot pressor") so we know when we've made contact. We also are working on "curb feelers" to automatically square up on the edge of the scoring platform to score, and programming an extensive set of "muscle memory" actions. The neat thing is that our robot will only have five actuators -- three for the H-drive, one for the lift motor, and one for the lift pawl (or possibly a brake). Over our short history, we've found that our success is inversely proportional to the number of actuators on the robot, mostly because it means more driver practice.