Cooperative Localization in Harsh Underwater Environment Based on the MC-ANFIS

Abstract

In this paper, a new cooperative localization (CL) method for multiple autonomous underwater vehicles (AUVs) is proposed to address the problem of measurement outliers and communication packet loss caused by the harsh underwater environment. Combining the advantages of both the maximum correntropy criterion (MCC) and the adaptive neuro-fuzzy inference system (ANFIS), the quality of collected data can be improved by MCC and the ANFIS can be better trained. The efficacy of the proposed method in the CL of AUVs is verified by lake trial. The experimental results show that ANFIS can effectively obtain the location of AUVs based on the input data when the communication packet is lost and the combination of MCC and ANFIS provides better positioning accuracy and robustness. When the probability of measurement outliers is 2%, the proposed method reduces the averaged localization error by 80%, the standard deviation by 84%, and the maximum error difference by 73% compared with the CL method based on cubature Kalman filter(CKF). Finally, the effectiveness of this method is verified by various experiments under different measurement outliers probabilities.