[oscully]

You are right that abort to surface is non-trivial.

I didn't work on Icefin, but I did write the comms-lost procedure (among many other things) on a previous ROV/AUV that was deployed in Antarctica to go under the Ross ice shelf. Britney Schmidt (one of the investigators quoted in the article) was the PI of the grant for that project as well.

Our AUV was deployed through a hole drilled in the sea ice and controlled via a fiber optic cable. It was used as an ROV in most circumstances, although we could preprogram an entire mission to be carried out autonomously. In the event of the cable breaking the submersible had to navigate via dead reckoning to get back within range of the acoustic beacon at the deployment spot. The AUV would follow the acoustic beacon until it was within visual range of a flashing led rod and then use computer vision to guide the AUV into docking on that rod so it could be recovered.

If an error occurred under sea ice then we could surface and bring out heavy equipment to drill a new hole to recover, but the ice shelf was too thick to drill through. We did have a magnetic beacon on the AUV that helped with location if we needed to recover after a failure under sea ice.