A method for co-evolving morphology and walking pattern of biped humanoid robot

We present a method for co-evolving structures and a controller of biped walking robots. Currently, biped walking humanoid robots are designed manually on trial-and-error basis. Although a certain control theory exists, such as zero moment point compensation, these theories assume humanoid robot morphology is given in advance. Thus, engineers have to design a control program for a priori designed morphology. It is very time consuming if both morphology and locomotion are considered simultaneously. Therefore, a method useful for designing the robot is proposed. First, simple models for both morphology and controller are used for the dynamic simulation, which are multi-link models as morphology and two kinds of controllers. One is a layered neural network and the other is a neural oscillator. Robots with optimal energy efficiency of walking are designed with the genetic algorithm. As a result, various combinations of morphologies and gaits are generated, and the relationship between the length of each link and moving distance is obtained, which gives an optimal energy efficiency. Moreover, the robots are encoded from the limited size of chromosomes