Technical Description
The morphologies of creatures and the neural systems for controlling their muscle forces are both generated automatically using genetic algorithms. Different fitness evaluation functions are used to direct simulated evolutions towards specific behaviors such as swimming, walking, jumping, and following. A genetic language is presented that uses nodes and connections as its primitive elements to represent directed graphs, which are used to describe both the morphology and the neural circuitry of these creatures. This genetic language defines a hyperspace containing an indefinite number of possible creatures with behaviors, and when it is searched using optimization techniques, a variety of successful and interesting locomotion strategies emerge, some of which would be difficult to invent or build by design.
The genetic algorithm is a form of artificial evolution, and is a commonly used method for optimization. A Darwinian "survival of the fittest" approach is employed to search for optima in large multidimensional spaces. Genetic algorithms permit virtual entities to be created without requiring an understanding of the procedures or parameters used to generate them. The measure of success, or fitness, of each individual can be calculated automatically, or it can instead be provided interactively by a user. Interactive evolution allows procedurally generated results to be explored by simply choosing those that are the most aesthetically desirable for each generation.
The user sacrifices some control when using these methods, especially when the fitness is procedurally defined. However, the potential gain in automating the creation of complexity can often compensate for this loss of control, and a higher level of user influence is still maintained by the fitness criteria specification.
Hardware / Software
Software and Animation: Karl Sims
Hardware: Connection Machine CM-5