AER-based robotic closed-loop control system

Abstract

Address-Event-Representation (AER) is an
\nasynchronous protocol for transferring the information of
\nspiking neuro-inspired systems. Actually AER systems are able
\nto see, to ear, to process information, and to learn. Regarding to
\nthe actuation step, the AER has been used for implementing
\nCentral Pattern Generator algorithms, but not for controlling
\nthe actuators in a closed-loop spike-based way. In this paper we
\nanalyze an AER based model for a real-time neuro-inspired
\nclosed-loop control system. We demonstrate it into a differential
\ncontrol system for a two-wheel vehicle using feedback AER
\ninformation. PFM modulation has been used to power the DC
\nmotors of the vehicle and translation into AER of encoder
\ninformation is also presented for the close-loop. A codesign
\nplatform (called AER-Robot), based into a Xilinx Spartan 3
\nFPGA and an 8051 USB microcontroller, with power stages for
\nfour DC motors has been used for the demonstrator.