Recognition systems data integration

As was mentioned above, joint operation of various recognition systems is used for detection of objects of different classes. Multi-beam echo-sounders and electromagnetic searchers can be used for objects with the significant sizes of cross section (pipelines). Here EMS can be used only for detection of metallic object presence in AUV vicinity, but not for finding its orientation. EMS and video camera can be used for objects with the small area of cross section (cables and thin pipelines). Let's further consider procedure of data processing and object position finding for these two cases.

Data integration of Electromagnetic searcher and Video imaging complex are executed for reliable AUV movement during communication lines inspection. Probability pt<n> is a factor of inspection object existence probability in time moment t regarding recognition system n. Accumulation of previous contacts can be made (fig. 6):

Where: pt<n>2 - total size of contacts by the moment of time t for a subsystem n; k<n> - factor of attenuation ("forgetting" information of subsystem n).

The general parameter of object presence pt22 according to all recognition systems is defined as the weighed sum of each subsystem estimations pt<n>2 and aprioristic estimation of object presence in the given point w0(XtAUV,YtAUV). Thus weights w<n> determine a degree of trust to each subsystem:

Average object direction and position are calculated on the basis of inspection object recognition systems data. The integrated estimation of cable or pipeline position and direction rtCAB=[XtCAB,YtCAB,®tCAB]T (fig. 6) are similarly calculated, including data rt<n>=[Xt<n>,Yt<n>,®t<n>]T of each subsystem n and aprioristic information about inspection object location in the given area r°(XtAUV,YtAUV) = [X°(XtAUV,YtAUV), Y°(XtAUV,YtAUV),

rt'">Z = k">• pr • rt<"> (17)

Where: rt<n>2=[Xt<n>2,Yt<n>2,Ot<n>2]T - temporary variable (a total component of each coordinate by the moment of time t for subsystem n). 