Low-Power, Bio-Inspired Time-Stamp-Based 2-D Optic Flow Sensor for Artificial Compound Eyes of Micro Air Vehicles

Abstract

We report a low-power, bio-inspired mixed-signal 2-D optic flow sensor to realize artificial compound eyes, which can provide a wide field of view (FoV) sensing capability for autonomous navigation of micro-air-vehicles (MAVs). Inspired by insect vision, a 2-D time-stamp algorithm has been developed to scale the number of deployable sensors at low power. The fabricated sensor estimates 16-b 2-D optic flows of maximum 1.96 rad/s with FoV of 60° from the integrated mixed-signal algorithm core, which consumes only 243.3 pJ/pixel or $\sim 30~\mu \text{W}$ at 120 fps. In addition, the peripheral circuits for modular design have been integrated on chip to provide optic flow data compression, wide-field integration (WFI), and serial peripheral interface (SPI). More than 25 sensors can be connected in a single SPI bus and transmit the full resolution optic flows. The fabricated prototype sensor supports full resolution 2-D optic flows from an array of $64\times 64$ pixels at 120 fps through a 3 MB/s SPI bus (4 wires total).