Abstract: To solve the problems of low precision and weak capability of resisting the noise disturbance caused by using traditional inertial navigation system/ultra short base line (INS/USBL) integrated navigation system, an INS/USBL nonlinear integrated navigation method based on observations of relative information is proposed. An observation equation is introduced based on difference between two incidence angles and between two slant ranges which are respectively resolved by INS and  USBL . Then, an adaptive noise statistics estimator designed for nonlinear systems is derived by applying the cubature rule based on the modified SageHusa algorithm. Simulation results show that, the positioning precision of the proposed algorithm is 2.4 times of that of the traditional algorithm. In the circumstance of unusual noise, the filter has better convergence and the system has better stability.