I'm about to start my thesis work on CS, and I'm pretty much set on the =
idea of society algorithms, more specifically ant algorithms. While =
initially I'll work only on the theory and creating a simulator to test =
the theories. (I already have a draft version of this simulator on my =
webpage: http://www.merlotti.com/EngHome/Computing/AntsSim/ants.htm )

All the concept is based on very simple individuals that when =
interacting in scale, produce interesting behavior. There are lots of =
research going on right now on ant algorithms for lots of different =
applications, including optimization algorithms like TSP (search for =
Marco Dorigo).

The next step is creating miniature robots that have very simple rules =
in it to try to simulate the ants behavior. I'm learning PIC, so I don't =
think that the software side of it should be a problem, basically the =
PIC will control movement and navigation. For the most part of the time, =
the movement follows a type of random pattern, except when they find =
pheromones.

In nature, pheromones are chemical drops that evaporate. Ants can detect =
the presence of pheromones through their antennas, and the fact that =
they do evaporate contributes to the creation of optimized nest-food =
trails.

Ok, now we get to the subjec of the message. If I want to implement such =
robots, how can I simulate the pheromone effect? So far I thought of 3 =
ways to do it:
1) Ant robots carry some kind of ink deposit with a drop control. This =
ink evaporates on time without leaving residues. Other ants will detect =
blobs of ink on the surface with a vision system. One of the drawbacks =
is that the environment must be controlled, so image segmentation and =
recognition doesn't get too complicated.
2) As in the first alternative, ants will carry some type of chemical =
substance with drop control, and other ants will "smell" that chemical =
concentration from the environment. This requires a chemical sensor in =
each ant robot, and I don't know how affordable it is.
3) Ants carry lots of heated marbles (will have to think on how to keep =
them heated) and they drop one at the time. Other ants follow those =
marbles by some kind of infra red detection. I saw some cheap robots =
that are attracted by heat, maybe this is the cheapest solution.

Although I have some budget to do it (personal), remember that the key =
concept here is population based solution, so 5 or 6 robots won't do it. =
(I don't know exactly how many, but I foresee at least 20).


Do you guys have any other suggestion regarding sensors, other =
techniques, anything?

Thanks

Padu
.--. .- -.. ..-
_______________________________________________
http://www.piclist.com
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist