Automatic path generation for group dance performance using a genetic algorithm

Abstract

Designing dancers’ paths in choreography, or a floor pattern, is one of the most highly creative tasks of choreographers. Aiming to assist this task, this paper presents a novel system that automatically generates a number of floor patterns for multiple dancers given a choreographer’s high-level feature inputs. The proposed floor pattern model represents locomotor movements of dancers on stage. Through a dance literature survey, four major features, i.e., time, space, symmetry, and entropy, were selected as feature inputs and mathematically modeled. My system uses a multi-objective genetic algorithm to achieve desired floor patterns given input features. It iterates from random floor patterns to the ones that satisfy users’ preferences while exploring the space of floor pattern with selection, mutation, crossover methods that are developed to fit the genotype of my system. User tests confirmed that my system generates a wide range of floor patterns according to user-specified input conditions. In addition, an actual dance piece was choreographed with the proposed method, which validated the usefulness of the proposed system. The proposed system is the first that automatically generates floor patterns for multiple dancers.