They just have or today are uploading new software to the rover to probably address the fears you have. I'm sure they're concerned that all bases are covered when moving the rover to the edge of a 200 foot deep crater. You can certainly see the way the edges have been pushed up by the impact that created it so you never know how the drifts might change as you get closer.

I have the feeling the biggest threat comes from getting stuck right close to the edge and having the rover try to extricate itself only to have it suddenly surge out of its sandtrap and plunge over the side.

I've thought often about how to write mobility software like that and the challenges are pretty steep. First, your indication of movement is by two references, the amount the wheels have turned (which is masked during slippage) and visual clues from the hazard avoidance cameras. In terrain like this, when everything around you looks the same and there's no rocks lying around to ping on, that becomes a bit tougher.

With Mars having no noticeable magnetic field, gyrocompasses, which can detect very tiny movements, don't work. No GPS is available. Some sort of radar system would be prohibitively weighty and complex. Maybe the ultimate solution is to put a big bicycle wheel on an arm off the back of the rover to act solely as a rolling odometer. The other option is to send rovers in pairs where each help the other navigate and can also send back wonderful shots of the other craft in action.

I would imagine future rovers, especially ones going to areas of deep sand or slippery traction are going to have much larger wheels much like dune buggies here on Earth have so it won't be nearly the issue.

Art