3 ms·
Not to be a downer, but I don't see that working to well for different shapes, only general sizes. 1. It would be nearly impossible to account for length of re
by tmh88j 15y ago
Not to be a downer, but I don't see that working to well for different shapes, only general sizes.
1. It would be nearly impossible to account for length of rectangular pieces. Every rectangular piece of same width x height could go through that hole vertically
2. Smaller pieces could fall through every hole larger than itself. So, unless you're using a multi-tier system it's pointless.
Basically, I think it could be easily done for general sizes of pieces, not shapes. I'd love someone to prove me wrong because that would be an awesome tool to have.
- seclorum 15y agoWell, first thing I thought of doing is getting all pieces in our mindstorm kit measured, so I have a grand "list of sizes", then sorting the list along all axes, and creating some sort of curvature hash that can be applied such that certain 'sort patterns' will occur through the series across multiple axes, such that the smallest parts fall furthest. This hash could be N elements wide for both axes, where N is the # of potential sieves in the box. Then I would scatter the series so that per-sieve, the longest parts are eradicated/expelled in an early layer. The expel part happens because the sieves are not laid flat in the box, but staggered/staged in order to utilize gravity. Around the gravity staging, I would put the N element filters in a way such that the outer edges of the box naturally catch the larger pieces and hold them closer to the layer exit. I guess I should just dump the parts on a scanner and get on with it, but it intrigues me that among the HN crowd, someone might have a much better idea or reason to pursue the electro-mechanical route.