CNN based central pattern generators with sensory feedback

In this paper the topic of including feedback from sensors in the central pattern generator (CPG) for a hexapod robot realized through cellular neural networks (CNNs) is addressed. An approach based on local bifurcation of the CNN cells constituting the sub-units of the CPG network is introduced, allowing control of the direction of the robot. Suitable control can be realized by changing the value of the bias of the CNN cells. Moreover, inspired by the idea of Braitenberg creatures, purely reactive control of the hexapod direction is illustrated with an example of a robot able to avoid obstacles.