Artificial Life: Term Project  
Introduction Previous Works Implementation/Results Conclusion Impressum References
  Karl Sims - Peter Eggenberger - Josh Bongard  

Previous Works:

In this section we present a composition of other major projects allocated in the same field - morphogenesis. We are mentioning some that are also dealing with genetic algorithm including the developmental phase, which means that genotype is not equal to phenotype. Where appropriate, the different projects are compared to each other and to our own.

Karl Sims, 1994, GenArts, Inc. in Cambridge, Massachusetts (details)
Peter Eggenberger, 1996/1997, University of Zurich (details)
Josh Bongard, 2002, University of Zurich (details)

Some problems, all of them had to deal with, are: How are the genomes represented and encoded? Is it a binary string or a valued string? Another question that all these researchers have come across is: How to select the individuals? Whereas Sims uses truncation, Bongard selects his individuals by elitism. There are also many different ways of doing mutation and crossover. In binary strings you can just flip a bit, in real strings you can change a digit. Crossover can be done by producing either the same length of crossover or a different one.

What they all have in common is that they use evolution to achieve solutions that human beings did not think of.