The issue of the space environment is an important one. True. There are
also issues of a purely AI nature. I have what could perhaps be
described as bees in my bonnet on this one.

I feel that in many ways robotics research has taken a wrong turn.
Robotics should be concerned about Engineering Physics, Geometry and
CAD/CAM and understanding these things in a generic way. Instead some
AI researchers have attempted to build into robots a quality of
"cuteness" rather than trying to get thenm to do useful things such as
assembling a CAD defined flatpack for example. Assembling an assembly
from parts and sub assemblies is a critical step.

If a program developed this understanding the task of ruggedizing it
for aerospace, while not trivial, would at least be a finite task.

Balance, going up stairs bipedially, is a matter of transfer functions
and also geometry. Leaving space aside think of the gaing obtainable by
replacing wheelchairs with robotic limbs that would allow bipedal
movement on stairs and rough surfaces.