Integration of multi-objective and interactive genetic algorithms and its application to animation design

Interactive GA (IGA) is one of the methods of solving decision-making problems when the computer cannot evaluate solutions directly. The method forces the users to evaluate all the solutions generated by the computer. Hence it is difficult to solve practical problems only using IGA because the user must evaluate a great number of solutions, and more sophisticated assistance by computer is needed. The paper considers problems whose solutions can be evaluated by the computer partially. We propose an IGA to reduce the number of evaluations by the users, enhanced by techniques such as multi-objective optimization and clustering. The proposed method is applied to a problem of generating animation of a pass-motion by hands so as to confirm usefulness of the method. Results of the experiments show that the proposed method can generate high quality solution with fewer stresses on the users.