The dancing genome project: generation of a human-computer choreography using a genetic algorithm

In this paper, we present an interactive genetic algorithm for the generation of human-computer choreography, using motion capture technology. First, we introduce the four steps of the algorithm to (1) define a movement vocabulary, (2) initialize movement sequences, (3) generate mutants, and (4) select mutant sequences to create a choreography. Then, we show how this approach is implemented in real time to create interaction among dancers. Finally, we run simulations to assess the convergence rate of the algorithm, before generating a simple duet for actual and virtual dancers.