OK. I'll throw together some kind of display from the bits and pieces I have laying around in my inbox.

The AUSVI machines are typically neutral-boyancy (within a couple of m of the surface) power-driven things — no ballast tank, and no worries about surfacing if falling below the limit depth.

One of the recent UAV comps has apparently (I not entuirely sure of this) involved getting directions for navigation by reading submerged signs, but other than that the very murky (typical) conditions in the test sites usually limits the use of any kind of camera or even lights.

But some kind of high-end (or hacked up) sonar is needed to perform several of the usual comp tasks.

Of course, wild creativity can sometimes subst for expensive h/w like synethic aperture side scanning device.

I have 1/2-produced a "camera bot" for some diving buddies.

Since it had to follow an audio beacon as well as carry a $30k camera, the brief included all manner of means to not lose or recover said equipment in case of equipment failure. (E.g. onboard air tanks and lifting balloons).

The resulting amateurish mess of wiring, mod-ed dive torches, cheap air values, hoses (equipment had not been purchased originally with a standard port diameter in mind :} ), tanks, and whatnot strapped to the typical plumber's nightmare frame made the prototype a "bit" of a monster, and nowhere near reliable.

While the basic machine was dived in a couple m of freshwater (and there were several nasty problems even then) the threat of having to eventually submerge the thing to 10-15 m in seawater stopped much more progress.

I've managed to hack together some very primitive sonar equipment from basic transducers and s/w running on underpowered platforms.

Automomously analysing it (to me) has turned out a bit nastier than vision processing.