Code center > Darwinbots3

bot sensing abilities...

(1/2) > >>

jknilinux:
Hi everyone, how's thanksgiving been?

I have an idea- we should restrict each cell to only sense one thing at a time. For example, if a bot is able, it should only have a 1x1 resolution view of it's environment, with no distance information. Also, if it can see, it cannot detect chemicals in it's E-grid, and vice-versa. If any other senses are implemented, they will also be mutually exclusive with all other senses, except for homeostasis-like ones. This both emulates real life AND fosters the development of Interesting-Behavior- and multi- bots. For example, to judge distance, we will need to implement at least two bots to use binocular vision.

In fact, this might have caused the cambrian explosion- once animals developed multicellularity, they could take advantage of many new senses, but at the cost of being really complex. From there, it avalanched.

What do you think?

Numsgil:
Why only one sense at a time?

Anyway, I have thought about not providing any distance information with sight, but eventually decided against it.  While it would be cool to have an environment where bots need to have binocular vision to judge distance, it's needlessly complicated and actually not much related to reality.  Animals with non binocular vision can actually judge distances reasonably well just from environmental cues, not the least of which is the focal length of the iris needed to get a clear picture.  You can try it yourself by closing one eye and putting your hand in front of your open eye.  You can tell your hand is close to your face because to focus on it everything else gets fuzzy, and if you look at something far away your hand gets fuzzy.

As near as I can tell binocular vision simply enhances distance resolution, especially with objects that are moving and doubly especially with objects with no environmental cues (a pop fly in baseball).

And ultimately, in the end, the environment should be just as amenable to single bots as to multi bots.  Otherwise having multibots is no great achievement.

jknilinux:

--- Quote from: Numsgil ---Why only one sense at a time?
--- End quote ---

Well, because it mirrors nature- you can't be single celled while judging distance and having vision. That requires multicellularity. If we could have remained single-celled and have had color vision and a brain, then we probably would. Perhaps a cell can have more than one sense, for a cost, like 0.1 bp per cycle for vision. So, to have multiple costless senses, you should be a multibot.

And maybe, reduce the resolutions bots have to only one pixel- aside from being totally impossible to make an eye out of a single cell with a resolution of 5 in nature, it costs nothing extra for single bots, and should encourage complex eyes in multibots to increase visual accuity.

Ta-183:

--- Quote from: jknilinux ---And maybe, reduce the resolutions bots have to only one pixel- aside from being totally impossible to make an eye out of a single cell with a resolution of 5 in nature, it costs nothing extra for single bots, and should encourage complex eyes in multibots to increase visual accuity.
--- End quote ---
Once that becomes feasible, of course. It would take weeks, if not months of development to get an MB like that to function.

Numsgil:
Keep in mind bots aren't necessarily single celled.  They're abstract enough that they could be cells, ants, lions, or whatever.

Navigation

[0] Message Index

[#] Next page