This page captures the navigation processing status and notional plan as of July 2009. It was written by John Graybeal based on meetings with Mike McCann and Rob McEwen.

The automated processing described at BI AUV Data Processing produces a gridded netCDF (.nc) navigation file based on the raw on-board (real-time) navigation. This follows a similar flow to the AUVCTD, and the operational procedures are described at the bottom of the file. They produce netCDF files that are visible to SSDS and accessible on the web.

The first-level strategy discussed with Rob is to perform post-processing in the same way as the AUVCTD, namely to correct the locations based on fixes obtained from GPS. In the case of BIAUV, the fixes occur at the beginning and end of runs, rather than at the top of each yo-yo maneuver, but the effect would still work. Mathematically, the error discovered at any fix is averaged into preceding positions, using linear weighting (vs any previous fix).

The idea for the software per discussions with Mike M is to copy the AUVCTD code, then modify it for this purpose. The copy has been made and lives in the biauv CVS repository at biauv/src/matlab. (omatlab is junk and can be ignored.) There are a number of products that the AUVCTD matlab files create that are not useful for the BIAUV; these parts of the matlab could be commented out.

Note that running this software is a manual process on the AUVCTD, requiring adaptations for each mission, and presumably would be the same for BIAUV.

The BIAUV has 2 additional navigation sources that are not available with the AUV CTD. The first of these is position corrections from sonar on the ship (telemetered acoustically); the second is position corrections from a target beacon (which presumably has been surveyed in to some known accuracy). Each of these could be used to improve the navigation, either in real time or in post processing.

In either case, some thought is required, as neither navigation system is as solid as GPS. Whereas a GPS navigation fix is guaranteed accurate within a meter or less, the fixes from the ship and beacon are subject to considerable noise. You do not want to "improve" the nav by fixing it to a bad/noisy position, especially if you might get multiple noisy positions in a row. (There is no Kalman or other filtering in the AUVCTD nav software.)

So the next steps for integrating these other fixes into the real-time and post-processed navigation will require some mathematical analysis to create the appropriate algorithms for possibly noisy fixes. (That was as far as the project got, last I was aware.)