[JesseK]

We did a curved rectangle suction test using thundercord as the seal, a pneumatic tube hard mount as the port to draw air out, and a pneumatic cylinder as the vacuum generator. It worked really well with human hands. It lifted that RC up quite nicely, even when the vacuum was placed at the very bottom of the RC and rotated up from a laying down position.

We then started to work on what it would take to get a good seal around a recycle bin every time with a robot. It is these auxiliary components which start to make everything complicated, since the two basic requirements of the overall mechanism aren't met without some serious extra 'stuff'. All told, it would take 8 buttons just to drive this mechanism, and it isn't even the primary mechanism (ironically that one only takes 2 buttons...).

We haven't written the concept off yet, but we did do a pros & cons list. Most of the cons are about complexity in one dimension or another, which tells me the design is very likely to be unreliable if we try it. We have some other designs which use some features that are similar to the auxiliary components of the suction design, so we're moving forward with detailing the little things out first.

If you do decide to go in the suction device direction, keep in mind the simple movements of the robot, and the workload you'll put on your driver.

The only advice I'll give outside of that is to not use suction for totes unless you have great drivers or a great alignment/counter force mechanism.