An FPGA-Based Image Recognition with Remote Update Functions for Autonomous Driving on “ad-refkit”

Abstract

This paper explains the ongoing development of an FPGA-based image recognition system for an autonomous driving robot car for the FPT'21 Design Competition. We are developing an FPGA-based image recognition on “ad-refkit” equipped with a Zybo-Z7 board to realize two main functionalities: 1) to provide recognition results in the actual environment of the contest through the Internet and 2) to be updated remotely based on the results. With these functionalities, we can realize high-performance and low-power systems using an FPGA.