Self-position Estimation of Small Fish Robot Based on Visual Information from Camera

摘要

Autonomously working small fish robots are ideal for surveys of fish resources and underwater structural inspections where electric wires hinder the movement of the robot or where radio signals, such as GPS, cannot be received. For practical usage, the fish robot needs to estimate its self-position to swim autonomously in water. Using visual information from images taken from the bottom facing camera, self-position estimation becomes possible. Digital image correlation and tracking method is often used to measure deformation or strain of material and visualization of flow in water. Although self-position estimation of aquatic robots becomes possible when this method is applied, microcomputers (PIC, H8, AVR etc.) cannot be used for the digital image correlation method since they are unable to perform the calculations in real-time. In this study, self-position estimation was tested utilizing the real-time digital image correlation method by using a FPGA (Field Programmable Gate Array). VHDL programming for FPGA and the FPGA equipped circuit board have been originally developed. This FPGA board executes both the control of the swimming fish robot and image processing at the same time. Results confirm that the small fish robot can measure swimming distance by itself.