3 ms·
For the obstacle avoidance mechanism, which they say realistically can only have one "subroutine", I was thinking it would benefit from a random perturbation to
by alanbernstein 5y ago
For the obstacle avoidance mechanism, which they say realistically can only have one "subroutine", I was thinking it would benefit from a random perturbation to the turning angle, to avoid getting stuck in a loop. But I wonder how you might implement a pseudorandom or chaotic movement in clockwork?
Maybe by summing via multiple weirdly-shaped cams? Or somehow extracting information from the motion of a double-pendulum system?
Impractical, perhaps, but I can't help wondering the best way to do such a thing.
- LadyCailin 5y agoIf you could somehow manage to track and maintain the movement of a triple pendulum, you can have basically true random input. https://www.youtube.com/watch?v=dDU2JsgLpm4 https://www.youtube.com/watch?v=dDU2JsgLpm4
- nradov 5y agoThe atmospheric wind is a pretty good source of randomness so attach a small windmill or windvane to the mechanism.
- spiritplumber 5y agohttps://www.reddit.com/r/mildlyinfuriating/comments/dzrgnk/roomba_with_no_vacuum/ https://www.reddit.com/r/mildlyinfuriating/comments/dzrgnk/r... This one works surprisingly well. I got one at a garage sale for $10 for a lark, and its fully mechanical "navigation" system is surprisingly reliable, in that it rarely ever gets stuck.
- spiritplumber 5y agohttps://www.reddit.com/r/crappyoffbrands/comments/dzs7ho/sweep_the_roomba_knockoff_in_action_by_request/ https://www.reddit.com/r/crappyoffbrands/comments/dzs7ho/swe... Something like this then