Intelligent Detection for Artificial Lateral Line of Bio-Inspired Robotic Fish Using EMD and SVMs

Abstract

In this paper, in order to solve the problem of accurately detecting the underwater moving target for the bioinspired robotic fish in the complicated natural interferences, where the acoustic perception in information interaction and perception system is vulnerable to multipath effect, and the optical sensing is easily affected by water turbidity, a novel intelligent detection method of artificial lateral line (ALL) is proposed. Firstly, these different intrinsic mode functions (IMFs) are decomposed via empirical mode decomposition (EMD) from original signals received by the ALL system, to separate the target signal from the interferences. Secondly, the characteristic frequency of the vibrating target representing the different fish is obtained from IMFs to detect the underwater moving target. Finally, these power spectrums of IMFs are input into the support vector machines (SVMs) to recognize the orientation of underwater moving target intelligently. Through tank tests, this method is applied to the ALL detection, and these testing results show that the proposed method has better detection performance than the traditional method, such as FFT and neural network, which provides a new way for the collaboration robot fish school.