Robot Navigation Using Neuro-electronic Hybrid Systems

Abstract

Neuro-electronic hybrid systems have been gaining interest of researchers as a possible architecture for computing. This aims to exploit the strengths of biological neuronal systems with their immense parallel processing and learning capabilities along with that of VLSI systems. Towards this end, we have set up a system which demonstrates the use of a live neuronal culture to solve a real world problem of controlling a robot doing the task of obstacle avoidance. We show that a neuronal culture can look at the sensor inputs to the robot and generate motor commands to allow it to explore an arena while avoiding obstacles.