Complex behaviour from simple scripts

ok, before I move on to the next phase of development I'd like to test out a few shoals of fish out in the "real world", ie. not under my constant supervision.

Can any members offer some water where I could put out a feeder and a few fishies?

One feeder supports about 10-12 fish. if the feeder is removed the fish will starve. If the fish swim away they will starve. If they over breed they will starve. They seem to be quite stable and sim safe... but I need to know for sure.

Offers?

but was just reading hamlet's article on your fish last night - if any of you haven't caught that, here it is:
http://secondlife.blogs.com/nwn/2005/06/evolving_nemo.html

fffffascinating stuff... as far as ai vs. non-ai approach, both have their purposes - a good ai would certainly be a much more processor-intensive approach for little fish; vicevers, the latter's program would be too simple for a robot (though, in an evolutionary sense, could very well be part of a logic plug-in for a larger ai).

you said you're using the same prog in the predator? so the predator can also be hunted? ("there's always a bigger fish" - quijon would be interesting to make a whole plethora of species to watch...

Hi,

Im new to SL but have been reading about these AL experiments. I was wondering if the parisite created by Icon could be used by the fish created by Surina. In stead of using AV's for transportation the parisites could use the preditor fish.

The food chain would then go as such: Preditor eats clown fish, clown fish eats parisite, parisite feeds on preditor. When attached to the preditor, the parisite could drain a given amount of life (amount of life drained by the parisites could even be used as variables (like with the fish's greed factor). After detaching they could reproduce.

It looks as if the clown fish would have difficulty attacking a small food source, and the parisites created by Icon change their size. It could be set that the parisites are only large just before reproduction, and the amout of times they reproduce could be allowed to mutate, making reproduction both favorable and non favorable.

I know these are two different projects, but I think their combination could yield some very interesting results. Poputlation growth would be much more dynamic (though I'm not sure how it would be controled). Also, it would create a fully functional food chain.

As I said, I am new to SL and I don't know how to script yet. I am just making a suggestion, and I already find all of these AL experiments (including the frogs thoroughly impressive.

Surina,

I live on a small island in Umber. I don't own enough of the surrounding water to provide a good place, but I do have a fairly large spa. Not sure how much room a shoal would require, but you can find the spa at Umber (182,103,20).

Would dearly love to have a shoal, which would enable me to spend hours watching them and waste even more time in second life!

Selador, please feel free to spend hours watching them on one of my islands. Bring a few friends, bring a lot of friends...

(and thanks for the offer... heading over to have a look)

Damn this is an interesting read. I've gotta head over and check it out soon

An interesting idea. I'll have to go back over my parasites then and check it out... seems my sky-jellies are going to be a while in the making.. (I want to make giant jelly-fish that float in the wind... but the tentacles seem to be a bit of a hurdle).

The fish now contain a list of different food sources. When a fish is born it randomly selects a food type to look for. If that food is not found it randomly chooses different food to look for untill food is sensed.

Different foods provide different amounts of 'nourishment'.

Alife projects combined

The fish can eat the frog eggs produced by Briggi Bards Alife Frog project...


Next step
Through some learning process yet to be devised, the fish will develope "prefered" food and pass this information to their offspring, (as greed is currently passed), eventually to the exclusion of some of the food sources. It is hoped that with enough choice of foods, seperate spiecies will evolve that eat different things.

The fish can now learn what food is available.

Every time a fish eats it remembers the type of food and it is added to a list. When choosing randomly from the list, if one food has been eaten more than others it stands a greater chance of being chosen.

The current food choice is reflected in the colour of their tail, giving the impression of several shoals each moving indepedently looking for different food sources. In reality, an individual fish will move from one shoal to another if it fails to find the particular food it wants and goes looking for something else.

I'm currently working on a way to encode the "food memory" to a colour vector so that the fish colour refelects it's overall food preference.



(I have also made modifications to a seperate part of the fish system which has to remain secret for now, other than to say it will soon be possible to go fishing...)

As always, astounding work.

I've spent too much time fish watching of late <g> and noticed some strange behaviors. A fish with a blue tail suddenly stopped in mid-water and turned upside down. A few seconds later a green-tail attached itself to the first and also went immobile, or so I thought. After a moment, I noticed that the tail colors were changing without any food input. Sometimes they synchronized, most times not. Some fish occasionally stopped and "watched" them for a second or two before swimming away. I did not see any fish bump them. After about five minutes the fish realigned themselves, one belly up, the other belly down. After another while, the first fish disappeared entirely. Minutes later, so too did the second. The tails changed the entire time.

Is this the result of a new feature, a server error, or what?

Fascinating.

The colour changing tails are easy to explain. It simply shows which food they are thinking about at the time. If they don't find that food they'll look for different food, hence the colour change.

As for going belly up, I've only ever seen that when trying to make a direct copy of a fish, using "Shift-drag", and I've always put it down to a physics error. They have rotation about X and Y axes turned off to keep them upright. I've always assumed that a rounding error or something during the creation process may cause them to appear slightly "leaning" and confuse the physics.


For you to see one go belly up on its own is new to me and I'd have to assume a physics error somewhere. It would eventually die off as it starved to death.

Hmm, this was definately a belly up move, but the fascinating part was the second fish attaching itself to the first while no other fish did so. In fact, most avoided the pair by a wide measure, even the other fish ran into me sometimes. Have you been fooling with genetic algorithms to create emergent behaviors?

Second note, why are the fish only found in the water? Why not in the air as well?

err... ummm... because they're fish?

Fascinating. I just finally popped over there and sat watching them for a while

What! You've never heard of flying fish?

This experiment actually made it into a local PC magazine which had a story about SL..

What? Way Kewl

Any chance of posting it here? What did they say?

Second Life got 84% and these comments:

IT'S: > Sociable > Creative > A bit rude in places
IT'S NOT: > Simple > Easy to drive > Really a 'game'

From "PC Games Addict" magazine.

The magazine has a forum.

They printed my name

I saw this on the news earlier and read up on it. I thought of you Surina

http://news.bbc.co.uk/1/hi/sci/tech/4313266.stm

Video here -

http://cswww.essex.ac.uk/staff/hhu/video/RoboticFishProject2005.wmv

Okay, that's freakin' cool! I've seen other animal robotics simulations before, but none that succeeded on that level of realism.

Come to think of it...I'm fairly familiar with the parts and technology they're using to build those...I might just have to get a bigger fish tank...*starts flipping through her unibomber catalog for parts*

The clips on TV were even better, they had a few of them all in the same tank. Some time down the line I bet people will have robotic aquariums in their living rooms. There would be a hidden 'recharge' bay hidden under some rocks where they swim to when power is low

Not to be outdone by a bunch of uninerds, I ran a population test today. While it is quite difficult to count a shoal of moving fish (even while 'frozen' with the edit tools they were still reproducing...) we agreed a consensus of about 190 fish.

During this excercise I discovered a new emergent behaviour that was hilarious to watch... to get the numbers up I set a single feeder to give food every 5 seconds. The fish quickly learned about this feeder and chose it as their prefered food, shoaling around it constantly, darting in each time more food appeared. Like a pack of hungry sharks. As the numbers in the shoal increased with the high availability of food, the circling mass getting ever more dense and impressive, it actually became easier to entrap them all with a shift-drag of the edit selector. 190 fish...

But of course, even while in edit time does not stand still, and so the longer they were held, the hungrier they became. The feeder was still producing food every 5 seconds and after a while a small 'cloud' of food prims was sitting in the center of this great shoal of starving fish. Time to let go of them...

WHOOSH, en mass they all headed for the food like a pack of hungry hounds at feeding time. What a sight. This is where I wish SL had a properly working video recorder.

You could work another type of fish, a small shark. If you made it work in packs, when they are near to the shoal they circle them, making the shoal take a giant circling action (as they try to get away from the preditors but the preditors are too close to each other for them to risk runnning). In this mode the non preditors should stop thinking about food at this point though (to avoid random darting out). The preditors should start darting through the pack picking them off once a certain time has passed. The inside of thier mouth should kill anything that touches it, everytime something is killed by it it gains health. The preditors will die once they have eaten all the no preditors though, so they should act as a population control.