Space & Astronomy/Space and Astronomy (October 1993).iso/mac/TEXT/SPACEDIG/V11_2/V11_224.TXT

Internet Message Format | 1991-07-08 | 17KB

Return-path: <[e-mail]>
X-Andrew-Authenticated-as: 7997;andrew.cmu.edu;Ted Anderson
Received: from beak.andrew.cmu.edu via trymail for +dist+/afs/andrew.cmu.edu/usr11/tm2b/space/[e-mail] (->+dist+/afs/andrew.cmu.edu/usr11/tm2b/space/space.dl) (->ota+space.digests)
          ID </afs/andrew.cmu.edu/usr1/ota/Mailbox/oa72YSW00VcJM:fU5r>;
          Fri,  6 Apr 90 01:29:03 -0400 (EDT)
Message-ID: <[e-mail]>
Reply-To: [e-mail]
From: [e-mail]
To: [e-mail]
Date: Fri,  6 Apr 90 01:28:32 -0400 (EDT)
Subject: SPACE Digest V11 #224

SPACE Digest                                     Volume 11 : Issue 224

Today's Topics:
           Space-tech excerpt:  Robots on the moon
----------------------------------------------------------------------

Date: Thu, 5 Apr 1990 20:20-EDT
From: [e-mail]
Subject: Space-tech excerpt:  Robots on the moon

This is the last of my recent batch of excerpts from the space-tech
mailing list.  Cheers!

 ///////////////////////////////////////////////////////////////////////
/// Marc Ringuette /// Carnegie Mellon University, Comp. Sci. Dept. ///
\\\ [e-mail] \\\ Pittsburgh, PA 15213.  Phone [phone](w) \\\
 \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\


Space-tech excerpt:  Robots on the moon     [330 lines, Mar. '90]

------------------------------

From: [e-mail]

I was just thinking of trying to sketch out an idea:  a robotic, teleoperated
"construction team" of about 50 little (say, 20-pound) robots designed and
operated in completely distinct ways (for redundancy and flexibility), which
would operate on the moon and be able to put together payloads shipped there
in kit form.  They would be teleoperated from an earth station, and use
common communications equipment....

------------------------------

From: Gord Deinstadt <geovision![e-mail]>

Marc, I have been thinking along similar lines for some time.  My
ultimate goal was a system of tiny robots that could assemble the
next generation of larger robots, and so on until they are big
enough to do "real" work.  But, even if they were less capable
they could still do a lot.

The biggest problem I see is power.  A 20-pound robot can't store much
energy relative to its size, at least not with batteries.  I'm not
sure how constraining this is, since a lot of the work can be done
while standing still, which makes a power cord acceptable.  You still
have to have a lunar power station, though.  Hmm- microwave transmitters
a few thousand miles apart, locked in phase, should be able to beam
power from Earth to a smallish rectenna on the Moon.

Anyhow, I'm convinced you can do a whole lot with fairly stupid robots
on the Moon, provided you're willing to do it slowly.  But slowly is
relative; we could put up a bunch of robots next year, so they'd have
a big head start over a manned base.  In fact I think this is a great
way to build the infrastructure for a manned base, not to mention
useful chores such as assaying minerals.

------------------------------

From: Joe Beckenbach <[e-mail]>

> The biggest problem I see is power.  A 20-pound robot can't store much
> energy relative to its size, at least not with batteries.

    Why not simply let each be solar-powered?  True, a little additional
stuff will be necessary to allow the 'bots to turn off at nearly the same
time, but that could be done in a small plain area by having each one
turn off when the sunshine level drops below a critical value.  So think
of it as a 50% duty cycle through each month.  1/3 :-)

    Put up instruments, a couple of small propulsion experiments, a
commsat or two, and a batch of robots all on one bus.  Put this in orbit
around Luna, and then send each item on its way, like blowing dandelion
seeds.

------------------------------

From: [e-mail] (George Ott)

How large and intelligent does a robot have to be to be useful for setting up
a lunar base?  Would a semi-intelligent bulldozer do the trick or do we need
something with some sort of arms?  What are the early tasks that need to be
accomplished?  Or that could accomplished using "off the shelf robots" or at
least robots built from off the shelf parts....
  One bobcat
  One Macintosh Si
  One satellite dish (for comm.)
  One SNAP generator for power
  Three grey scale CCD cameras for seeing

------------------------------

From: [e-mail] (Henry Spencer)

Remember that you need either (a) plenty of power to run heaters, (b) a
heated garage, or (c) robots built to soak at cryogenic temperatures
during the lunar night.  The Surveyor landers tried to survive using
small battery-powered heaters in crucial places plus durable hardware;
not all of them made it through one lunar night, and I don't think
any made it through two.  The Lunokhods and the Apollo instrument
networks did better, but they had isotope power.

If I were trying to use off-the-shelf hardware, I think I'd opt for the
heated garage.

------------------------------

From: [e-mail] (George Herbert)

Besides using a non-space rated brain [a mac SE???]
that robot would be using a thirty watt RTG [oy] for
power.  Do you know how little you can do with thirty watts?

------------------------------

From: [e-mail] (George Ott)

Some mention has been made of running the lunar bobcat on solar power.
What are the advantages of solar power over radioisotope power in this case?
Which has a higher power per lb. launched from Earth?
How do the systems compare with regards to reliability?

------------------------------

From: Gord Deinstadt <geovision![e-mail]>

My guess is that you could manage about 100 grams/sq metre for lunar
solar cells (thin-film type).  This yields on the order of 100 watts/sq
metre, or 1 kilowatt/kilogram.  I don't know the numbers for radioistope
thermal generators, but they are considerably worse.  An RTG needs a
radiator plus shielding, both of which are relatively massy.

Space probe designers use solar cells whenever possible, falling back
on RTGs when the probe is going to the outer solar system.

Both solar cells and RTGs have a serious problem when used
on the Moon; dust!  Every moving vehicle is going to stir up
dust, and that is going to coat any nearby solar cells or RTG radiators.
For the RTG, we can just wipe off the radiator every once in a while.
For solar cells, we have to worry about scratching the surface.
The obvious solution is to put a hard coating on the
solar cells (say, diamond film) but this is unlikely to have the
optical properties we need for best efficiency.

If we want to use solar cells (which I believe we do, based on power
density), we may have to locate them in a suburb far away from the
dust of the lunar base.  And we'll still have to worry about dust
thrown up high by rockets taking off and landing; maybe the only
way to escape this is to build a hard-surface pad.  Or employ a lot of
robots wiping solar panels.

Now they're stupid, shivering robots doing menial labour! :-)

------------------------------

From: [e-mail]

If you go to the North pole of the moon you can get a 6 month day.
Solar power will work well even at the low angle because there is no
atmosphere.  You can drive around for 6 months.  Then getting, off to a
timed and planned start, you can spend 14 to 28 days migrating south
for the winter (curve around as you go south).  The moon is small
enough, and there are no fences, so far.  Net result is that we could
drive a remote control bobcat year round.  I think of it more as a
remote control space buggy than a bobcat, since it will probably be
very light and take awhile to push dirt around - but hey, we will have
the time.  If you have two of these with some remote control arms such
that you can use one to work on the other you may even be able to keep
them both working for some time.

    -- Vince

------------------------------

From: "Edward V. Wright" <[e-mail]>

Jerry Pournelle's Lunar Society spent considerable time working
on a lunar bulldozer very much like the proposed Bobcat.  They
called it the Go-cart.  It was an ATV-type vehicle with a waldo
on the front.  It was designed to be teleoperated from Earth:
autonomous most of the time, but call for help when you encounter
something that you don't understand.

As far as migrating from one pole to another, I think you may be
overestimating the distance a vehicle can cover, over unexplored
terrain, in a day.  The Moon is about 6000 miles in circumference,
we're talking a 3000-mile trek, so 100 to 200 miles per day. That
doesn't sound very fast, but as I recall the proposed Martian rovers
cover only a fraction of that distance in a day.  You could
probably improve on that some, but only at the risk of losing
vehicles. (I'd plan on some attrition, anyway. There are a *lot*
of holes on the Moon.)

It might be better just to land two groups of robots, one at either
pole.  Operate each group of robots for six months at a time.  Or
build a tower that's tall enough for its top to be in sunlight all
the time.  The robots spend 6 months of the year roaming free and
6 months tethered to the base of the tower, doing work around the
base camp and getting their power from extension cords.

------------------------------

From: Gord Deinstadt <geovision![e-mail]>

I see a need for three types of robots.
1.  A truck (solar powered, see above).
2.  A backhoe (more flexible than a bulldozer - used for all kinds of
    heavy lifting and moving).
3.  A "monkey", ie. a small thing with high-dexterity effectors for doing
    fiddly tasks like repairing other robots.

>Personally, I favor the "semi-autonomous" theory of operation, where if the
>fairly stupid robot checks in with his mentors whenever he encounters anything
>beyond his scope of "reasoning."

I hate to see projects stalled because we're too ambitious in
our software designs.  I don't know what AI is capable of these
days, but for Lunar work and a short design cycle I'd even settle for simple
remote control, which is why I said the robots could be slow.  Semi-
autonomous would be better if it is available more or less off-the-shelf.

------------------------------

From: [e-mail]

I'd like to say more about my vision of a Robot Workshop, and along the way
make some comments about the discussion so far.

My goal is to extend a virtual human presence into an area the size of,
say, a large room:  the Robot Workshop.  In orbit, this might be a way
to do Space Station assembly; on the moon, it could be the means of
constructing and maintaining a small mineral processing plant or mass
driver.

I prefer an octopus configuration, with one central power/communications
system, plus ten or twenty small tethered robots.  This has major
implications for keeping each robot simple:  they need be no more complex
than robots in a lab on Earth, with power and control connections, motors,
cables, and little else.  Preferably, they would have some decent force or
touch sensors to provide enough information to the operators.  I think it is
essential that each robot NOT be required to have its own power source, radio
link, processing unit, etc.

As I envision it, the central system has solar cells or an RTG, communications
with Earth including several video channels and teleoperation control
channels, and hardware to control which of the robots is being used and
which video channels to ship back.  I see the robots operating in a 95% to
100% teleoperated mode.  My reading on the usefulness and flexibility of AI
techniques for operating the robots is that for the various tasks you'd run
into in a workshop, a skilled human operator can beat AI almost all the time,
even with a several-second round trip delay.  The only roles I would look for
software to play is are to monitor for problems and do an 'emergency freeze'
if an exceptional condition occurs, and to provide local control of simple
short-duration operations such as sanding or polishing.  I'm pretty solid
in this conviction, being an AI guy myself.  If you CAN teleoperate, you
SHOULD teleoperate.

What kind of robots would we need?  I imagine anchored robot arms, arms on
tracks, arms on wheels, arms with snap-on hand attachments such as wrenches
and screwdrivers, cameras on arms, cameras on tracks, fully mobile dune
buggies with clips for attaching payloads or cords for hauling, arms with
three-finger hands, arms with grippers, arms with hooks, toolkits, glue,
duct tape...

How do we use these most effectively?  I suggest a team of very talented
operators, each with separate controls, but working in a common command
center where they can yell at each other and help each other out.  Not all
the robots would be operating at once, particularly if the video bandwidth is
a severely limiting factor, which I believe it might.  There should be
mock-ups of the environment so they can work out problems in the mock-up if
they need to before trying it in real life.

=============

What does my proposal provide?

  - A simple, flexible way to project human presence into a workshop.  It is
    relatively low-tech and straightforward.

  - A heterogeneous and flexible system with minimal hardware requirements.

  - A system able to do simple assembly and repair.

=============

On to some specifics people have mentioned:

On temperature control:  I bet using heaters really isn't that expensive.
Isn't a vacuum bottle what you keep things hot in?  Maybe we can get some
figures from outer-solar-system probes.  I figure all we have to do is make
sure that the heat flow is good enough that no part gets left in the cold.
Maybe lighting the area with some major floodlights would help.  And rather
than using electric heaters, perhaps small RTG pellets used for pure heat,
rather than power generation.

On self-reproducing robots: no way, no time soon, and CERTAINLY not with lunar
materials!  It would be nice to have robots that could do simple repairs
on each other, and standard parts would be nice, but that's about it.

Heterogeneity is a big goal.  I would really like the robots to be able to be
developed and operated completely separately, for instance by university
projects in several places.  Coordination is a BIG hassle, so let's just skip
it by making the interface simple - control lines in, video out, and each
project can do its own thing.  Hedges your bets, too.

Gord Deinstadt is definitely right that we should expect things to happen
VERY SLOWLY.  I think that's OK.  There are a lot of hours in a month.

On migrating south for the winter:  definitely not - you only move across
terrain if you have to.  Not to mention that my proposed system can't move
at all.

On RTG's versus solar cells:  I think solar wins by a lot, in terms of
weight, but RTG's provide power 100% of the time.  I don't know which wins:
if power isn't the dominant weight requirement, it might be worth getting a
100% duty cycle by using RTG's.

I don't mean to ignore suggestions for, say, a mobile, independently powered
bulldozer.  I think it's a fine idea.  However, I think a lot of the power of
my proposal comes from the fact that a simple system of video cameras and
teleoperated arms is very easy and very useful.  Amortizing the cost of the
rest of the system -- power, communications -- over a number of cameras and
arms makes them a lot cheaper.

------------------------------

From: Tom Neff <[e-mail]>

Two refinements to the Robot Workshop:

 - You'd want at least two separate power busses and radio circuits
available in the workshop (even if both are not constantly in use), so
that repairs to a failure in either one can be done using the other,
without automatically incurring a human repair call.

 - Although teleoperation beats AI at Earth-Moon distances for many
things, I think it would make a lot of sense to give the robots enough
sense to obey "GRASP X", "PUT A ON B", "UNSCREW" and so forth.  The
feedback loop would tend to interfere with smooth operation otherwise.
And it is something we can do, unlike "FIGURE OUT WHAT IS WRONG AND FIX
IT AND REPORT BACK" as some of our more eager beavers would go for.

 - How about five arms and three eyes per station.  The teleoperator
picks an eye-pair for stereo vision, but can switch pairs at will for
a change in perspective without having to move or wait (also an assistant
can monitor his work from the side).  The five (or whatever) arms
include a "bottom arm" with a support platform attached to hold small
work, two general purpose manipulator arms and two hold/transport arms.

==

This concept could work in orbit or on a planetary/satellite surface.
In all cases you need a reliable automated way of getting new stuff
TO the workshop.  In orbit the Progress system works.  On the Moon
you have other problems.  An automated soft landing + breakdown
technique, combined with a teleoperated 'drag rover' might work.
But this won't be simple to achieve.

------------------------------
[ End of excerpt ]

------------------------------

End of SPACE Digest V11 #224
*******************