Virtual pet progress

In the labs today I was amused to glimpse a dragon on the island next to a squirrel. There was an acorn beside the dragon labelled "squirrel bate". I began to have a bad feeling about the poor squirrel, and further investigation revealed that -indeed- the squirrel is programmed to be lured towards the acorn where it will get crunched up by the dragon. When I went back to get a screen shot, the squirrel was gone. Eaten? Does the dragon look guilty?

This example of virtual nature's cruelty was inspired by a class I did on artificial life (boids and such) where I was trying to inspire them to set up eco systems on the island, and explain the concept of emergence.  You can see a video of my example cookie monster mating behaviour here: http://www.macs.hw.ac.uk/~judy/InteractiveSystems/cookimonster.avi.  You can see the slides:

Download Class9.ppt if you are interested in the code. It is a very cool example of emergent behaviour - the monsters unexpectedly go mental and have loads of babies all in a row. It was a surprise to me when it happened when I was recording the screen cast.

At the moment the students are trying to get their pets' tails to wag. This turns out to be ridiculously hard in Second Life and requires more maths than is good for me (or the students). I had to come clean the other day and admit that I was still researching how to fix a problem with code to do rotations of linked prims. (If anyone know how to do this, let me know). There is a silver lining effect which is that three students have researched this for themselves and come up with three different but viable solutions which they have agreed to share with the class. There are some very nice examples of peer collaboration, self motivated research and applying knowledge from classes to new situations here. I will write these up a vignettes sometime, but in the mean time I have a frustrating weekend of trying to make a cat's tail twitch.

Comments

[Kobolds]

I liek this concept of creature reproduction and setting up virtual ecosystems, but surely they are only viable if the offspring grows into the adult versions, and elderly (uneaten) animals perish from age. Would it not be possible to implement this thanks to the state based nature of linden script?

I'm probably thinking too basic here but defining different states based on emotion/situation and age. As they passes out of their default infant stage they grow in size and develop new behaviour.

With a week left I doubt I have the time to implement this but just some musings that have occurred to me.

[Judy]

Absolutely you can implement that. :-) There are two directions you could take this 1) genetic algorithms for controlling the evolutionary fitness of the creatures. 2)Modelling emotions and pesonality. Either of these would be current research topics in the department and you will learn about them in third year. Assuming you are one of our students...

[Kobolds]

I would be very interested in finding out more about the research the university is doing for both of these.

Sadly though due to having three courseworks to do other than the IS coursework I won't have time during this term. I'd be interested to continue learning about both genetic programing and simulating emotional responses.

I know that you and Sandy are working on the emotional line of research, who is working with the genetic programing?

p.s You don't happen to be looking for students to help with the research over summer? :P

[Judy Robertson]

Ruth Aylett or maybe Dave Corne can direct you to genetic algorithms people. We did actually have some summer research student internships last year. Keep a look out for the adverts around May!

[Betty]

I recently came across your blog and have been reading along. I thought I would leave my first comment. I don't know what to say except that I have enjoyed reading. Nice blog. I will keep visiting this blog very often.

Betty

http://adoptpet.info