The objective of this project is to study evolution process in a computer-simulated artificial world. We hope that, like in the real world, despite randomness and aimlessness of basic evolution mechanisms, it will lead to creation of more and more efficient "artificial organisms", still better and better adapted to the "artificial world" conditions.
For many years people have used computers to simulate the nature. This field of science has been named "Artificial Life". The name is obviously connected with "Artificial Intelligence" - both fields of study partly cover. AL, compared to AI, goes more with biology and physics, and tends to be more practical than theoretical. It might be called a branch of biology, as we study living (or "living", depending on the definition of this word) organisms in an environment. No matter that the environment is an artificial world inside a computer: philosophers haven't yet decided whether our world is real and, nevertheless, biologists keep examining its living organisms. In addition, our made-up worlds let us (unless we have visited other very distant planets) study signs of life which has nothing to do with proteins.
Scientists studying AL concentrate on various fields and serve many purposes. Rules of simulated worlds do not have to be like real ones, but such models seem more interesting (perhaps because you are able to directly compare simulation results with reality).
Boids are an example of creatures following few simple rules, which anyway make them group with others (like fish shoals or bird herds). This kind of behavior was used in film making (computer control of animals' models - "Lion King", "Batman forever" and many advertisements). More sophisticated models include learning and self-improving creatures. Artificial organisms (i.e. carriages with engines) can be learnt to avoid obstacles, seek energy spots, follow targets, escape enemies etc. Such experiments are directly connected with real robots control - they might alone perform useful actions.
Often it is evolution which controls improvements - it awards better adapted (or, in the case of artificial world, efficient as we want them) organisms. That has to do with genetic algorithms and works the same way. Of course, considering robotics, we cannot (still?) speak of creating new "organs" nor new robot species as results of mutations or crossovers.
For the evolution process to go so far it is required that organisms should be entirely under control of the evolution. In the real world it is possible because of nature having "discovered" genes as a way of formal description of construction plan. Such descriptions can be then mutated and crossed over.
Such statements are true with our world simulator, where genes describe the structure of the organism. Full freedom in creating genotypes means theoretically ability for creation of creatures of any complexity.
Such experiments make us ask whether self-improvement of artificial organisms can lead to founding consciousness, intelligence, feelings? Not in our simple model, of course. But evolution has once showed it is up to that. We had better watch out!