NavList:

All my experience is below sea level (except when friends and I got bored and borrowed the bridge's sextant of course-- the 2nd and 3rd mates were happy for the practice, Masters' were good).  I can't speak to Sabertooth and am certainly no submariner, but AUV navigation is literally the dayjob... all this computer-free CelNav stuff with tables and slide rules seems like black magic.

I've found AUV navigation to be little-changed in principle from techniques familiar to any traditional navigator.  For a Sabertooth-- or any other deepwater AUV-- as Howard says primary navigation will be through a Dopper Velocity Log (DVL)-aided INS.  The Saab website lists a IxBlue PHINS as standard-- we use them too, do recommend.  While no model was specified, those are all strapdown Fiber Optic Gyroscope (FOG) systems with-- for comparison-- a standard alignment time of 5 minutes by the clock.  Near the seafloor, this 21st-century INS "solution" still relies heavily on those DVL velocity measurements.  Nearly the same as an aircraft's Doppler Rader system.  The embedded computer may be the envy of a proper sailor-- it applies a ground-velocity update every 0.1-2 seconds-- but in principle the whole operation is not so different from Bowditch's Putnam approaching Salem harbor in dense fog.  Even a gravely-mistreated FOG+DVL can typically manage error less than 0.5% or so of distance travelled; I'd dearly love to put any sailor's DR stats up for comparison when giving a talk. Without a DVL-- i.e., on descent-- most AUV INSs are exactly as accurate as those in civilian airliners.  That 3km trip from the surface presents quite the challenge, as the INS will drift civillian-aircraft-fast until DVL's max range of perhaps a few hundred meters (mostly 50-200ish m, the latest long-range RDI systems ~500m).

Most folks resolve this dilema with some type of acoustic navigation system.  In the Wedell sea, anyone sensible would use an Ultra-Short BaseLine (USBL) system entirely contained on the ship.  Sonardyne's AvTrak 6 can be combined with their Ranger II topside system to get range/bearing/elevation+communications.  It's bad form to second-guess folks in the field-- they probably used what they had on-hand-- but if they couldn't fit a Ranger II head on Agulhas II's drop keel they could certainly fit one through the moon pool on a pole (seriously-- how cool is that ship?  A moon pool for in-ice operations!?!).  Combined with the ship's GPS+INS... really not so different from a stadiometric range + bearing to a lighthouse but in 3D, yes?  If asked I'd eschew any seafloor transponders.  All the rage pre-2000, today's USBL+GPS+INS systems let the intrepid underwater navigator skip lengthy deployment, survey, and recovery processes.  It's dang hard to get anything to sink/float faster than ~90m/min (I've tried).  At the Endurance site a seafloor system would need at least 30 minutes down for each of at least 3 transponders, plus an hour or two to survey in each transponder, plus 30 minutes each for them to reach the surface-- and no guarantee that surface would be ice-free.  At probably >$30,000 per transponder and order($1000's/hr) for the ship noone will want to run that risk-- heck, I once screwed it up without ice.  Not to mention all the time required to move the ship (again, possibly through ice) to transponder locations for both launch and recovery.  No thanks.  You may read about other underwater navigation methods-- chip-scale atomic clocks are becomming common in oceanographic stuff these days, and the old-timers like their pseudoranges-clocks-be-darned-- but relying on these adds money and engineering risk and time and especially bunks.  Critically, a USBL system tells the ship how to find the AUV.  Effort is usually needed to forward that position to the AUV-- otherwise the AUV's operators on the surface ship might know the robot's lat/lon better than the robot itself does.  Seafloor transponder systems... often the robot knows position better than the operators.  Can be a real bear if you lose comms and the AUV can't tell you where it thinks it is-- and you almost always lose comms before ship-to-vehicle nav.

And that's what keeps this relevant to traditional navigation-- the core concepts haven't changed.  I use the running fix to teach modern INS/Kalman-filtering concepts to anyone who isn't the interested kind of Engineer.  It's clear only a fool believes the position plotted by (a $literally-realestate-worth$) of space-age nav instruments-- a Proper Operator reads between the fixes, understands the error modes.  Maybe doesn't always get it right because there's other stuff going on, more at stake-- but gets the AUV home safe regardless.  And if that takes plotting range rings in pencil, in 2022, in the post- RDF age of AIS, Iridium, GPS, whatever-- so be it.  I've seen the pencil happen (ok, way back in 2021).  Sure, the stakes are lower-- its only an (insured) robot-- but when out of options you still pray.  Maybe a different kettle of fish, but still a kettle, full of seacritters. Surely air navigators said something similar circa 1922.

I happen to have some experience in the scientific AUV world through the US National Deep Submergence Facility (AUV Sentry) and recently other systems, but I'm sure any oil-and-gas folks here would merely laugh at how simplistic (and underfunded) this view is.  Still, might help color the picture-- but its a small world, anyone with specific details should please butt in!

--Ian

PS If anyone wants to do any deepwater search, I happen to know of a 6km AUV with time to spare (i.e., not Sentry). To save time: Gary makes a better case that TIGHAR.