We propose a method to acquire a hierarchical behavior controller which is achieved by shifting phase between local behaviors. Complex behavior is hardly obtained using any unsupervised leaning methods, because of enormous search space. In order to reduce the search space, a hierarchical behavior structure is effective. We have previously proposed a hierarchical behavior controller, which consists of two kinds of modules: behavior coordinator and behavior controller. It is applied to the control problem of a seven-link brachiation robot, which moves dynamically from branch to branch like gibbon swinging its body. Numerical simulations demonstrate that the obtained controller can successfully achieve continuous locomotion.
展开▼