AUV positioning and motion parameter identification based on observations with random delays

A stochastic observation system model with random time delays between an arriving observation and the factual state of a moving object is adapted to identify its motion parameters. Equations for optimal Bayesian identification are given. A conditionally minimax nonlinear filter (CMNF) is applied to solve the problem in practice. The design procedure of the CMNF, including the choice of the filter structure, is discussed in detail on an example of autonomous underwater vehicle (AUV) positioning based on observations of stationary acoustic beacons. A computational experiment is carried out on a model close to practical needs using three variants of the filter, namely, the typical approximation of the updating process, the method of linear pseudomeasurements, and the geometric interpretation of angular measurements.