Driving Hexapods Through Insect Brain.

Insects are really astonishing creatures if their learning and adaptation capabilities are considered. In this paper two specific characteristics of their tiny brain are taken into account: classification and sequence learning. A complex neural network architecture, refined in the last few years, is reviewed and applied to a neuro-inspired controller for a hexapodal robotic structure. Classification is performed using Morris Lecar neurons, that receive input stimuli from a lattice of spiking neurons, arranged similarly to the insect Mushroom Bodies neuropile. A specific network devoted to context formation is used to recall the learned sequences and to retain relevant subsequences. A typical example of a labyrinth solution through the capability of learning and recalling sequences of visual objects is dealt with. The computational model is reviewed, refined and experiments on a hexapodal structure are discussed.