THE BLIND NEURAL NETWORK MAKER: CAN WE USE CONSTRAINED EMBRYOLOGIES TO DESIGN ANIMAT NERVOUS SYSTEMS?

The exploitation of a few simple principles (symmetry, segmentation, and recursion) has enabled the economical description of a wide variety of natural forms by the parametrisation of the underlying growth rules. The application of these principles to the specification of neural networks is investigated, and expressions derived for the consequent savings in complexity. The issues raised are discussed in the context of a neurally-controlled fourlegged robot.