A Combined Indoor Self-positioning Method for Robotic Fish Based on Multi-sensor Fusion

Abstract

In an experimental environment with limited conditions, it is always hard to achieve precise positioning of robotic fish. A combined indoor self-positioning method in this paper is introduced to solve the problem. For the short-distance range, coordinates are calculated by fusing the measured distances and angles. For the medium-distance range, a clustering-grid supervision (CGS) algorithm is proposed and adopted to correct the coordinates obtained by the four-point positioning method. An ostracion-like robotic fish is used as the experimental object to achieve centimeter-level positioning with an average positioning error of 4.492 cm in a short-distance range and decimeter-level positioning with an error of 2.049 dm in a medium-distance range. Compared with traditional methods, this comprehensive method has the advantages of low cost and high accuracy.