Phase oscillator neural network as artificial central pattern generator for robots

We design an artificial central pattern generator and we exemplify its integration using different prototypic virtual robot models. The artificial central pattern generator is modeled through a network of phase oscillators. Virtual robots are controlled by this central pattern generator using an interface that interpolates stored angular states of their target poses. Finally, we consider control signals coming from the robot's environment to the central pattern generator and to the interface between the central pattern generator and the robot. We show that the control signals can change the dynamics of the system in order to modify the robot's movement patterns. We study three cases: changing the frequency of the pattern sequences, switching between gaits of locomotion and the reorientation of a robotic entity.