Providing a good way of testing out a man-tended space habitat.
One of the ideas NASA is looking at incorporating into ARM is attaching a prototype deep-space habitat. This would allow visits of up to 60-day duration by crews of up to four. While there are other ways you could test something like this (such as Lagrangian point L1/L2 gateways), testing it in an operational environment would be useful, as would demonstrating the ability to do long-term habitation in close proximity to an asteroid.
Now, there are good reasons for testing habitations at Lagrange points, I grant you. For one thing, being able to establish the means of setting up waystations along a path or set of paths to service a greater exploration and exploitation of our solar system is invaluable. Having a predictable nexus for inbound/outbound spacecraft is better for a the kinds of manufacturing, materials processing, and even passenger-handling systems that have proven so effective for the past couple of centuries. And such waypoints’ efficiencies will be highly optimized by the systems themselves determining where they reside in space, not merely making do with what exists out in the heavens.
However, Goff is absolutely correct that testing habitat technologies would be well-served by setting up shop on asteroids first, for a couple of reasons:
Firstly, we don’t want to be endlessly floating in space watching as we glide past or around the objects we really want to set foot upon. Landing a habitat on an asteroid satisfies at least that human urge to explore. But, more than that, it allows a much smaller effort there to be extrapolated to more demanding landing and living missions to places like Mars, and we can learn a lot more with less fuss. A big driver in Mars landings is that we have a much more difficult time achieving safe entry and landing speeds from the planetary approach, as we lose most of the atmospheric braking advantage we enjoy landing on Earth. To fix that, we generally either have enormous parachute size (and mass) to catch what little air there is on Mars, or we carry even more enormous amounts of mass in the form of landing propellant. Airbag landings, while shown feasible for our robotic missions, are not reasonably safe enough or predictable enough to land human systems, up to and including the humans themselves.
In contrast, recent missions such as the Dawn orbiter mission to the dwarf planet Ceres and the Rosetta mission to Comet 67P show that we can indeed orbit exceedingly small objects in space where asteroids reside, and can even land on them. The propellant mass to perform those landings is quite small.
Secondly, by setting up crew-tended shop on asteroids, you can tackle more than one thing at a time. You can credibly test the habitat while the crew actually performs the needed inaugural experimentation needed to assay an asteroid. You end up learning more than one thing at a time that way.
So, until issues with landing downmass and technology for more demanding crewed missions are worked out, the habitats-on-asteroids mission profile allows for early-term validation of off-world living while concurrently pursuing industrial research.