Research On Leader-Follower AUV Cooperative Position Algorithm

Nowadays,autonomous underwater vehicles(AUV)play an important role in military reconnaissance and surveillance,anti-submarine and patrol,marine surveying and mapping,marine resources exploration,diving support and other fields.When AUV can not continuously obtain the GPS information of the outside world,no Mothership tracking,and can not obtain its position information through terrain matching information or gravity under certain circumstances,it will lead to the loss of navigation ability.In view of the shortcomings of the above positioning methods,the information sharing between the AUVs makes the positioning accuracy of all AUVs in the team improved and the error is bounded.This method is called cooperative localization and has gradually become the mainstream research direction of underwater AUV positioning.In view of the shortcomings of the above positioning methods,the information sharing between the AUVs makes the positioning accuracy of all AUVs in the team improved and the error is bounded.This method is called cooperative localization and has gradually become the mainstream research direction of underwater AUV positioning.In this paper,a cooperative localization algorithm based on extended Kalman filter is proposed in this paper.The distance and velocity between AUVs are used as the measurement values.It improves the positioning accuracy from AUV and achieves the same order of accuracy.Aiming at the problem that the underwater observation information is not enough and the measurement data update rate is low,which directly leads to the weak observability of the system,the observability conditions of the positioning system are theoretically analyzed.The matrix condition number theory is used to quantitatively analyze and simulate the observability conditions of the system,which provides an effective basis for designing a reasonable coordinated positioning scheme.The time-varying of underwater acoustic channel and the complexity of noise interference have a great influence on acoustic ranging,and abnormal results of measurement often appear.In order to improve the robustness of the algorithm to measurement noise,an adaptive anti-outlier filtering algorithm is proposed,which improves the positioning accuracy and robustness of the system.Firstly,in order to fully consider the prior information of noise,the attenuated memory filter method is used to estimate the adaptive noise.Secondly,the residual test method is used to detect the outliers and the gradual elimination index is used to correct the outliers by summing the residual weights.The results show that the adaptive anti-outlier filtering method can estimate the variance of ranging and velocity noise better,detect and eliminate outliers effectively,so as to obtain higher positioning accuracy.In addition,the performance of cooperative positioning is evaluated,and the influence of ranging error and velocity error on positioning accuracy is analyzed.The results show that the influence of ranging error on positioning results is far greater than that of velocity error.Therefore,in practical application,the emphasis should be placed on improving ranging accuracy.In order to develop the state estimation performance of the co-location algorithm under actual conditions,a coordinated positioning test system was constructed by using surface ship and underwater anchor beacons.Among them,the underwater anchor beacon is assumed to be a leader AUV with a known position,and the surface ship simulation is performed from the follower AUV,and the solution from the AUV position is completed by the acoustic ranging and the speed information fusion.Under the conventional algorithm,the positioning error is less than 1m in the case of single piloting,accounting for 10.4%;in the case of dual piloting,the positioning error is less than 1m,accounting for 59.7%;under the adaptive anti-field algorithm,the positioning error is less than 1m in the case of single navigation,accounting for 61.9%.In the case of double navigation,the positioning error is less than 1m and accounts for 85.1%.The experimental results show that the proposed algorithm can effectively improve the positioning accuracy and stability of the system compared with the conventional algorithm.