A Cellular Neural Network (CNN) is the core circuitry behind our pattern generation for bipedal locomotion. The following papers will help you understand the basics of a CNN:
Please note: the CNN is used to generate the patterns (walking, running etc) for locomotion. It does NOT do anything else (like guarantee dynamic stability). However, by using a CNN we have eliminated the need to generate a locomotion pattern. Once we decide on a suitable locomotion pattern, we can design a suitable control law.