Hi Kevin
Thanks for understanding where I am coming from you know I want you to succeed so I can piggy back off what you have done
Totally understand cutting it down to be brief as there is a lot that goes into a beast like this
Another (sort of) way to dissipate the torque induced on the main body of the unit is if you can make the rear fins somewhat adjustable (see pic similar to the ROV lights and then lock them off maybe ±5 degrees?) although the induced torque would be related to the speed you could set them once (like trim tabs on outboards) for the “average speed” the AUV would tend to run at so they counter act the induced torque and lower speed perhaps too much and a higher speed too little and then you would have to rely more on the distance between the CB and CG
Again maybe too little words for the skid steer (if you go with 2 motors) I would say a “turn” would be done more with the front lateral thruster but a minor course correction or “trim” with skid steer (think driving slightly into a off centerline current) for slight corrections (say a couple of degrees) given the lateral arm moment of force from the lateral thruster from the distance from the CG it would require very very little force(ie hard to control)[eg bow and stern thrusters are great for spinning a vessel on the spot but a really poor on course correction whilst underway]
PS bet you a beer there will end up being some sort of keel 
Great to hear on the endcap rigidity the Fishplate joiners from the Amethyst AUV is
As an assumption I would had assumed that the AUV would fly about 3-4m off the bottom given
- It’s a “lightweight” or “consumer grade” AUV [not sure of a good name ie not big and chunky with 24hr duration] I would assume it would be limited to say low range [swath] SSS in part due to their [SSS] power requirements and transducer size (think 900 – 1200kHz Trident Starfish or Deep Vision system running at high frequencies giving really great detailed images but poor [comparable] swath[ie not a wreck hunter but a site documenter]) and hence if running SSS you want it 10% of the SSS range off the bottom and the assumption it could most likely “see” the bottom at this height off it
- Its other most likely payload is photogrammetry and or video transits and thus also needs to see the bottom
(I would love some other “cheap” payloads (towed mag or fixed metal detector [there is a difference]) lidar, which other than MB they would all be pretty close to the seabed)
Given this (and it’s me just discussing it) I would have thought it would be an easier navigation route than a developing a low cost DVL (and its associated low cost accuracy limitations)
Again thanks and take it as it’s intended 
Scott