FPGA implementation of elementary motion detectors for the visual guidance of micro-air-vehicles

Abstract

Here we describe a field programmable gate array (FPGA) implementation of a bio-inspired visual sensor designed for estimating the optic flow onboard micro-air-vehicles (MAV) and improving the stabilization and visual guidance performances. Optic flow determination depends here on elementary motion detectors (EMDs), the principle of which was originally based on an electro-physiological analysis of the housefly's EMD neurons performed at our laboratory. FPGA technology appears to be well suited for use with several EMDs with visual fields covering various subregions of the MAVs fields of view. The multi-EMD architecture described here processes the signals emitted by a small linear array of photoreceptors operating in the current-integration mode. The experimental data presented confirm the validity of our multi-EMD implementation. The small 12/spl times/12 mm sized FPGA used here could incorporate up to 245 EMD.