Developing sensorimotor behavior with a network of spiking neurons

Artificial nervous systems have been realized with spiking neurons fabricated by VLSI. To explore ways of developing networks of such units through sensorimotor experience, a simulated vehicle, equipped with tactile and visual sensors was trained, using a biologically realistic Hebbian mechanism to navigate an obstacle course. Starting with a simple tactile reflex, networks developed visual guidance mechanisms based on the detection of luminance and motion.