Re: Beserk
The thinking was that as technology develops, certain tasks become easier. Some of it is also potentially solving other problems too. The current Mars rovers are all very limited in their navigation capabilities, they rely heavily on commands from Earth to do anything, which is partly why it takes them so long to get anywhere. One of the challenges in the competition was that of navigation; how reliably can you navigate when you don't have a compass (no decent magnetic field on Mars) or GPS, and on uncertain terrain so wheel odometry can't give a reliable indicator of distance?
The starting premise was that the "lander" containing the recovery robot would be in a known position, accurate to a few metres, based on being observed by passing satellites. The locations of the sample containers is known to a similar accuracy. Thus, the task is to navigate autonomously from the base to an approximate area and search, allowing for the fact that there may not be a sample in the area to find, if it's been buried. This is where it gets more interesting, because some teams used an approach that moved from sample location to sample location, collecting them all, before returning to base and disgorging the lot. Other teams worked on the principle that if a sample was found, return it to base immediately, which also gave opportunity to reset the navigation to a known starting point. The first approach clearly reduces total travel time, but is literally carrying all the eggs in one basket if the rover failed, whereas the latter would mean that if it failed after one or more successful recoveries, there would still be something worth sending back.
Also in there was object recognition, which was a lot less mature ten years ago - it was necessary to recognise the sample container in order to pick it up.
As for recovery, the plan was that the base station would be a rocket capable of reaching orbit, where it would rendezvous with an orbiter that was designed to reach Earth orbit, after which they could drop the sample package for retrieval in a manner that has since been demonstrated.
Given that they've got mobile rovers, adding an extra gadget to obtain and package samples is a relatively low incremental cost and risk, which is what's happening now. They may yet decide to leave them there if other technologies overtake it, but remember that this was long-range planning from many years back, trying to predict what advances would be made over the next 30 years or so.
From memory, over the five years they ran the competition, I think only three teams managed to get samples back to the platform under field conditions, and an impressive demonstration of the eggs-in-basket scenario when one of the otherwise more-successful robots decided to head off through the boundary fence instead of returning to base, due to a programming error. Some teams were university teams, some were one or two people doing it for fun, others were slightly larger groups.