Design of a novel Central Pattern Generator and the Hebbian Motion Learning

In this paper, we propose a new CPG model and hebbian learning rule for the CPG. The output of the proposed CPG is determined by only phase differences of synchronization of component oscillators. A phase synchronization can be regarded as an adaptive behavior to an environment, so the CPG has an adaptavility despite of only simple connections to the environment. We also propose a motion learning rule for the proposed CPG. Since the rule is described by only simple signal processing, so that it can be realized by electric circuits easily, it will be able to be used efficiently for high degree of freedom robots.