Automatic Integration Method for FPGA-Based ROS2 Nodes for Parallel Processing

Abstract

With the development and spread of robot technology, intelligent robots are expected to play an active role. To utilize these robots with low power and low latency, tools that automatically integrate FPGA (Field-Programmable Gate Array) into ROS systems have been actively studied. This enables low-power and high-speed processing of robotic applications on FPGAs. In this study, we developed meta-FOrEST, a tool to automatically include multiple FPGA logics in ROS2 nodes, with a view to utilizing FPGA in ROS2. In the evaluation, latency, message loss rate, and power consumption were measured using ROS2 nodes containing multiple FPGA logic generated by meta-FOrEST on FPGA board with a Zynq SoC. The results showed that FPGA processing using ROS2 with 8 nodes in parallel is feasible with low latency and low power consumption. Beyond robot applications, we also envision this approach’s potential use in MEC (Multi-access Edge Computing) scenarios in the future.