TABLE OF CONTENTS The role of feedback in the CPG for walking is essential to deal with complex environments. In this Chapter the topic of including feedback from sensors in the CNN-based CPG is addressed: an approach based on a local bifurcation of the CNN cells constituting the sub-units of the CPG network is introduced, allowing the direction of the robot to be controlled. The dynamics near the bifurcation is furthermore exploited to include feedback from ground contact signals in the CPG. Moreover, the speed of the pattern can be controlled in a very simple way by adopting a switched-capacitor technique to implement the CNN-based CPG. This leads to the so-called hybrid approach, in which the core of the control is analog and a digital control provides the signals modulating the outputs of the analog core. At the end of this Section experimental results obtained with the first prototype of the CNN-based CPG chip are discussed.
The approach described here aims to include feedback in the CNN-based CPG and relies on a local bifurcation of the CNN neurons constituting the sub-units of the CPG network. Suitable control can be achieved by changing the value of the bias of the CNN neurons, according to a strategy inspired by the idea of Braitenberg creatures [Arkin (1998); Braitenberg (1984)].
Braitenberg showed how surprisingly complex behavior (that he compared with cowardice, aggression, and other forms of human behavior) can be obtained by using relatively simple direct couplings between the sensors and motors of vehicles.
