Aiming at outliers existing in the underwater acoustic measurement in a cooperative localization(CL)system for multiple autonomous underwater vehicles(AUVs),a master-slave CL algorithm based on cognitive mechanism of hippocampal spatial cells is proposed.Firstly,grid cells are modeled by 2D continuous attractor network for path integration based on velocity and heading information of slave AUV.Secondly,multiple grid cell plates are self-organized using competitive Hebbian network,and a place cell plate with a single firing spike are generated based on multi-scale grid cell plate.Finally,the distance measurement between leader AUV and salve AUV is converted into cell firing activity and update place cells to correct the accumulated path integral error of slave AUV.Since place cells gradually approximate real position through multiple firing updates,the negative influence of measurement outliers can be effectively reduced.The simulation results indicate that the proposed brain-inspired CL algorithm is more robust to outliers existing in the measurement noise,and the localization accuracy can be improved by 28%compared to the threshold EKF-based CL algorithm.
维普
万方数据
