A modular FPGA-based implementation of the Unscented Kalman Filter

Abstract

Nanosatellites, while lowering the cost and ease of space access, suffer the issue of reduced performance compared to larger satellites, particularly when it comes to attitude determination and control. Field Programmable Gate Arrays (FPGAs) have been used in the past to make up for the shortfall in capability but tend to have more complicated development processes than general purpose microprocessors. To simplify development as well as promote portability and reusability between satellite missions, a hardware/software co-design of the Unscented Kalman Filter (UKF) implemented on a FPGA device is presented. The design is implemented on a Zynq-7000 XC7Z020 to establish proof-of-concept and verified using simulated data. The design achieved a 1.5× speed-up over a purely software implementation and the resource usage and power consumption are both low enough to be integrated into a full SoC.