3 ms·
I had considered, but not tried, using this algorithm for a dark-seeking robot (perhaps a pedigree of the Braitenberg or BEAM style robots). Knowing where the
by JKCalhoun 4y ago
I had considered, but not tried, using this algorithm for a dark-seeking robot (perhaps a pedigree of the Braitenberg or BEAM style robots).
Knowing where the darkest place in a room is seems unknowable for a simple robot.
So take a random walk for perhaps 3 minutes or so while recording the darkest light level. Then wander for another 7 minutes stopping if you record a lower light level during the initial wander.
Within 10 minutes I would like to think your robot will have done better than average at finding the darkest place in a room.
- syntaxing 4y agoIsn’t this a typical maze problem for a robot but the cost function is optimized for reward for darker spots? If that’s the case wouldn’t a-star hybrid or particle filters solve your problem?
- JKCalhoun 4y agoMaybe. I'll have to look into the algorithms you mentioned. To me there was a Secretary Problem use-case because we don't know what the light amplitude range is - we don't know what "dark" is (and will vary at different times of the day, vary by room, etc.). Also it's simple enough that you can imagine even a cellular-size organism implementing something like it.