🚀 go-pugleaf

In article <HAGERMAN.93Jan7224103@rx7.ece.cmu.edu> hagerman@ece.cmu.edu (John Hagerman) writes:
>
>There has been some discussion about the tradeoffs between using one
>large robot and many small robots for planetary exploration.  Has the
>Erebus experiment changed the opinion of anyone here?  I don't want to
>restrict attention to the specific results; for example, I don't think
>we learned anything new about fault tolerance.  Rather, I'm hoping for
>new thoughts engendered by considering the experiment in general.  In
>my mind, for example, the experiment supports large robots for such
>use since a fairly standard large robot design could be fairly easily
>modified to conceivably navigate the volcano, and it seems less likely
>that the same would hold for small robots.  Any other thoughts?

Actually, I came up with a completely different opinion based upon the
failure in Antarctica.  If anything, I believe this adventure has shown the
key weakness in the single robot proposals: if something goes wrong you are
completely screwed.  The scenario in Antarctica was similar to a Mars
exploration in that it is remote and a hostile environment, but one big
difference is that there will be no one around to crawl to the rim of a
canyon to reset the computers or perform other maintenance on Mars.  There
will be no modification of robot design once the mission to explore Mars
begins, you only get to use what you take with you.  Please remember that
the people paying for the exploration are not going to appreciate it if we
end up sending Mars it's biggest paperweight...

If anything, the failure of Erebus seems to point to the disadvantages of
using a single large robot to explore hostile, unknown environments.  With
a single robot you have a single point of failure for the entire mission,
if your robot ends up with machanical difficulties or other problems you
have just wasted millions.  With multiple smaller robots you still have
other robots to continue exploration and you have the option of making the
smaller robots capable of some specialization to explore unique
environments.

The Erebus experiment seemed to provide nothing to support the single robot
proposals, I am really curious as to how anyone could find support for them
under the circumstances of this situation...

jim
-- 
Jim McCoy          |  PGP 2.0 Key md5 hash:
j-mccoy@nwu.edu    |  61323c49024cb089337c78a05aafe8c6
mccoy@ils.nwu.edu  |  finger mccoy@ils.nwu.edu for public key

I believe that many small robots is the only viable method to explore
planets because it allows redundance as a feasible option.  The risks
involved in a single large robot are far to great to take considering
the numerous variables involved in planetary exploration.  As an example
NASA used TWO voyagers and TWO martian landers in the mid-seventies.  Numerous,

-------------------------------------------------------------------------
 Eric Paulos                    Robotics Laboratory
 paulos@robotics.berkeley.edu   University of California, Berkeley
 (510) 642-3248			Berkeley, CA 94720
-------------------------------------------------------------------------

In article <GERRY.93Jan8231255@onion.cmu.edu> gerry@cmu.edu (Gerry Roston) writes:
>The
>question of using small robots versus large ones is almost a religious
>question.  The greatest failing of those proposing small robots is the
>naive assumption that using multiple, small robots increases overall
>system reliability.  This is not true because one can not assume a
>priori that the failure modes are independent.  That is, if the robots
>are identical, and one of them fails in a particulr manner, this
>implies that the others will also be prone to that particular failure
>mode.

I'm impressed with what the Ambler group has accomplished, and very
impressed with what the Erebus group accomplished in a very short
amount of time.  I think they deserve a great deal of credit, despite
the unfortunate glitch that ended the test.

That having been said, I'll have to come down on the "small rover"
side of the debate.  (I'll admit to being biased, since I spent a
summer working with the JPL microrover group.)

The failure modes may not be independent, but at least the failure
occurrences are.  Suppose there is a 10% chance that a certain failure
will render a robot immobile sometime during its lifetime.  If you
have a single large rover, you have a 10% chance of losing the rover.
If you have ten microrovers, you will probably lose one, but you only
have a 1% chance of losing all ten of them.

The other advantage of multiple rovers is the ability to obtain data
from several sites at different locations on the surface of Mars.

>From a technical viewpoint, the major drawback to small robots in
>telemetry.  To transmit a signal from Mars, you need a moderately
>large antenna and a bunch of power, if you want reasonable data rates.

How about putting a satellite in Mars orbit and using it as a relay
between the rovers and Earth?  Would that solve this problem?

It seems to me that the primary disadvantage with small rovers is
their inability to carry large scientific payloads.  My opinion is
that this disadvantage is outweighed by the advantages above, but I
suppose it depends upon one's particular interests in research and
exploration.
--
_______________________________________________________________________________

Brian Yamauchi			Case Western Reserve University
yamauchi@alpha.ces.cwru.edu	Department of Computer Engineering and Science
_______________________________________________________________________________

By today's standards, the Viking landers are large craft.  The
question of using small robots versus large ones is almost a religious
question.  The greatest failing of those proposing small robots is the
naive assumption that using multiple, small robots increases overall
system reliability.  This is not true because one can not assume a
priori that the failure modes are independent.  That is, if the robots
are identical, and one of them fails in a particulr manner, this
implies that the others will also be prone to that particular failure
mode.

From a technical viewpoint, the major drawback to small robots in
telemetry.  To transmit a signal from Mars, you need a moderately
large antenna and a bunch of power, if you want reasonable data rates.
I do not have my references in front of me for doing link
calculations, but from the moon (which is right next door), you need a
.5 m dish on the moon, putting out 20 W of power to a 10 m dish on the
earth to get data rates of between 10k-100k bits per second.  The
problem from mars is much more difficult.  In addition, mars has winds
which will make antenna pointing difficult, thus degrading data
transmission.  To overcome these problems, a larger vehicle is needed.
By larger, I am thinking of a vehicle with a mass of 100-200 kg, as
opposed to a number of micro rover concepts that have a mass of < 10
kg.  These micro rovers would be incapable of meaningful scientific
exploration. 

--
Gerry Roston (gerry@cmu.edu)    | It would be almost unbelievable, if history
Field Robotics Center,          | did not record the tragic fact, that men have
Carnegie Mellon University      | gone to war and cut each other's throats
Pittsburgh, PA, 15213           | because they could not agree as to what was to
(412) 268-3856                  | become of them after their throats were cut...
                                | For some reason, too deep to fathom, men
The opinions expressed are mine | contend more furiously over the road to
and do not reflect the official | heaven, which they cannot see, than over their
position of CMU, FRC, RedZone,  | visible walks on earth.  Walter Stacey
or any other organization.      | 

The Erebus project was probably the single most ambitious robotic
undertaking ever attempted.  To have accomplished what the Erebus team
did was a minor miracle itself.  The failure of the Erebus project
says little to nothing about the one big robot versus multiple robot
scenario.  Had the project been less ambitious, i.e., had there been
more time available, the problems that beset the mission would never
have occured.  It's well and good for people sitting in their snug
offices running robotic simulations to make snide comments, but they
should actually try to develop a real system and see what happens.

Although I do not agree with all of the design decisions made by the
Erebus team, they achieved a goal that mo other organization in the
world would even dream of, and I am damned proud to be a part of this
organization.  This mission has brought robotics into the homes of
people around the world and has shown what robotic technologies are
capable of.  Even though this project did not achieve its stated
scientific goals, the mere fact that a viable, working system was
delivered, basically on time and within budget, will help future
generations of robotic researchers by providing a bench mark of what
can be done within our field.

--
Gerry Roston (gerry@cmu.edu)    | It would be almost unbelievable, if history
Field Robotics Center,          | did not record the tragic fact, that men have
Carnegie Mellon University      | gone to war and cut each other's throats
Pittsburgh, PA, 15213           | because they could not agree as to what was to
(412) 268-3856                  | become of them after their throats were cut...
                                | For some reason, too deep to fathom, men
The opinions expressed are mine | contend more furiously over the road to
and do not reflect the official | heaven, which they cannot see, than over their
position of CMU, FRC, RedZone,  | visible walks on earth.  Walter Stacey
or any other organization.      | 

In article <HAGERMAN.93Jan7224103@rx7.ece.cmu.edu> hagerman@ece.cmu.edu (John Hagerman) writes:
>There has been some discussion about the tradeoffs between using one
>large robot and many small robots for planetary exploration.  Has the
>Erebus experiment changed the opinion of anyone here? 

Well, I wouldn't say it changed my opinion, but I thought it sort of
showed an advantage of many small robots: if one bit of machinery fails
it will only stop one robot, not the whole project. I also couldn't help
thinking that small robots might have been easier to get to the crater
in the first place. But obviously in the absence of a sucessful
deployment of a robot herd for a comparable task (as yet) there's no 
proof that it's the better way to go.

BW

>>> On 9 Jan 93 10:13:13 GMT, yamauchi@ces.cwru.edu (Brian Yamauchi) said:


BY> I'm impressed with what the Ambler group has accomplished, and very
BY> impressed with what the Erebus group accomplished in a very short
BY> amount of time.  I think they deserve a great deal of credit, despite
BY> the unfortunate glitch that ended the test.

I'm sorry, I must have been napping during the Erebus accomplishment.
What did the trip to Erebus do that couldn't have been done on a 30
meter slag heap near Pittsburgh?  Teleoperation?  That could be done
with a programmable delay.

The truly impressive accomplishment is that the robot was built in one
year.  Everyone in the project deserves recognition for that.

But in my opinion, the choice of testing area and testing methodology
was incredibly poor, serving only as a media magnet.  Imagine a
biologist performing an experiment while juggling test tubes on a
trampoline.  If the experiment worked, it would be truly impressive.
Not because of the scientific results, but because the experimenter
overcame the completely unnecessary self-imposed barriers.





--
Mike Bolotski          MIT Artificial Intelligence Laboratory
misha@ai.mit.edu       Cambridge, MA 02139     (617) 253-8170

hagerman@ece.cmu.edu (John Hagerman) writes:
: So, my subject line got you to read at least this far :-]
: 
: There has been some discussion about the tradeoffs between using one
: large robot and many small robots for planetary exploration.  Has the
: Erebus experiment changed the opinion of anyone here?
: I'm hoping for
: new thoughts engendered by considering the experiment in general.

I've heard at least one argument for small robots in planetary explor-
ation, i.e. their small size making them easier to transport to the
planets surface, the 'redundant systems' effect of numerous small
machines, and the ability to cover more ground in a given time span.

The argument for large robots is that they can be more sophisticated,
and can traverse more difficult terrain than their smaller counterparts.

I wonder about 'blending' the two philosiphies... Imagine that you had
one large, agile, and reasonably intelligent robot, and say, six or so
smaller less sophisticated ones all linked together via some sort of
network (on a radio link maybe?). The smaller robots could act as remote
sencory input devices, or data collection drones, that are directed by
the larger one. The larger one ("Mother") would collect the data per-
ceived by the drones periodically, and transmit what was relivant back
to Earth. When a drone encountered an obstacle (such as a cavern), it
would inform the mother of the problem, and the mother would decide
whether to 'stay the cource' or not. If so, she would collect up her
drones (as a marsupial would) and attempt to negotiate the obstacle
herself. After the obsticle had been navigated, she would release
her drones to continue their work.

			. . . .

Of course, this is only a very broad idea I'm presenting, as I'm
just a guy that likes to dream at his keyboard :-)

: 
: - John
: --
: hagerman@ece.cmu.edu

Fred

-- 
Fred Floberg - N1GVA at Evil Weather Machines LTD.
email: ...!uunet!hsi!a3bee2!n1gva!fcf

Actually, as one of the people who was offered an opportunity to work
on the Erebus project ( I turned it down for personal reasons ) I
think I can offer more insight into the "failure" of the Dante
explorer robot.  

Due to what I see as political and organizational reasons, 
the project was supposed to proceed from start to finish in
one year.  Once the negotiation and contract signing stuff was taken
care of, that time line had been shrunk to just ten months.  Ten
months to take a VERY complex problem through all phases from design
to actual mission operation.  That's a hell of a short time even for
the fine folks at the FRC who are used to difficult deadlines.

One of the simplifing decisions that was made to help meet the
schedule was to exclude any forms of redundancy in the system that
would cope in the event of any single point failures such as the
fiber optic link failure that resulted in the mission being scrubbed.

In retrospect this may have been a "bad" decision but it seems that
without it there would have been no way to complete any portion of
the project in the time allotted.  In a real mission there would
almost certainly be a more generous time/money budget and a
careful evaluation of the possible single point failure modes would
have been made.  Ample redundancy would be designed into all mission
critical systems that would allow operation to continue under a
variety of failure scenarios.  

Overall, I think the Erebus project was a qualified success, in
that it must have been one of the most rapid development 
projects ever to have met with even partial success in 
such a hostile environment. Not just the final mission environment but
the pressured environment under which the development project was
carried out.

Hopefully, they will get another shot at it next "summer" with 
a machine that has had the advantage of an additional
year of unhurried development.  

My hat is off to the Erebus team who came through under intense
pressure ( and temperature 8^) ).
Cb

I'm really sorry not to add nothing meaningfull to this discussion but
I don't know what went wrong with the experiment.

   Could anyone sumarize it in a few words, please?

     Thanks 


     Jos'e Velez

In article <GERRY.93Jan8231255@onion.cmu.edu> 
	gerry@cmu.edu (Gerry Roston) writes:
>system reliability.  This is not true because one can not assume a
>priori that the failure modes are independent.  That is, if the robots
>are identical, and one of them fails in a particulr manner, this
>implies that the others will also be prone to that particular failure
>mode.

Unless the 'failure modes' are on the order of "I've fallen and I
can't get up!".  This also assumes all of the small machines are
identical and will haev the same failure mechanisms.  Multiple,
co-operative machines are more likely to be able to help each other
out of tight spots.  Unfortunately, multiple machines also seems to
imply smaller ones.

>From a technical viewpoint, the major drawback to small robots in
>telemetry.

This one is trivial to solve, the mother ship that lands the smaller
robots has the large antennas, high power amplifiers, and even a lot
of the 'brains' of the multiple smaller rovers in it's vicinity.

>These micro rovers would be incapable of meaningful scientific
>exploration. 

I think everyone is setting up straw robots for the other side's
viewpoint.  Possible rover sizes range from the CMU Ambler and the
Army's (future?) teleoperated bulldozers all the way down to the MIT
microbots and Drexler's nanomachines.  My own feeling is that a small
number (say 3 to 6) of medium sized (say 3-4 feet on a side) vehicles
can be very useful, but then that's the size vehicle I built for my
Simulated Lunar Teloperations projects, and the _major_ size
constraint on that was that it had to fit in the back of a VW Golf!

Willie Smith
wpns@pictel.com
-- 
Willie Smith     wpns@pictel.com     N1JBJ@amsat.org
"That's the wonderful thing about crayons, they can take
you to more places than a starship."  Guinan - STNG

>hagerman@ece.cmu.edu (John Hagerman) writes:
>: There has been some discussion about the tradeoffs between using one
>: large robot and many small robots for planetary exploration.  Has the
>: Erebus experiment changed the opinion of anyone here?

In article <1993Jan10.131539.21053@n1gva> fcf@n1gva (F. C. Floberg) writes:
>I wonder about 'blending' the two philosiphies... Imagine that you had
>one large, agile, and reasonably intelligent robot, and say, six or so
>smaller less sophisticated ones all linked together via some sort of
>network (on a radio link maybe?).

You have "hit the nail on the head". Each type of robotic system has
both advantages and disadvantages which might satisfy some mission
requirements but not others. This is an intelligent compromise.

Two years ago, I worked at the Space Exploration Initiative Planet 
Surface Systems office and performed feasibility studies about the 
kinds of robotic systems which were proposed for the Mars missions.
I had no vested interest in any specific technology, and was amased
when the technology brokers came racing out of the research centers 
and universities, trying to convince NASA to consider their robotic
systems. As you would expect, all of the systems we considered met 
some of the requirements, but no single system could meet all of the 
requirements. It makes sense to have several different robotic 
systems. 

The idea of multiple specialized robotic systems for Lunar/Mars
exploration is supported by two general guidelines: 1) decoupling,
so that if one system performing one part of the mission fails, it
does not prevent other systems from completing other mission goals,
and 2) redundancy, so that if one system fails, another systems can 
take its place.

Oh! I forgot about the fact that we need a really big budget, and a
launch vehicle capable of getting all this equipment there. Maybe
we can live with just one tiny robot ... 
:)


-- 
**************************************************************
Philip Lee Graves, Lockheed-ESC, Houston, TX        
graves@arabia.jsc.nasa.gov 
**************************************************************

In article <GERRY.93Jan8231255@onion.cmu.edu> gerry@cmu.edu (Gerry Roston) writes:
>By today's standards, the Viking landers are large craft.  The
>question of using small robots versus large ones is almost a religious
>question.  The greatest failing of those proposing small robots is the
>naive assumption that using multiple, small robots increases overall
>system reliability.  This is not true because one can not assume a
>priori that the failure modes are independent.  That is, if the robots
>are identical, and one of them fails in a particulr manner, this
>implies that the others will also be prone to that particular failure
>mode.

You know Jerry, I'm really happy when you shoot your mouth off like this
because it makes me look diplomatic by comparison.

That using multiple small robots increases realiability is not a "naive
assumption", it is a theoretically and empirically verifiable fact.  It
does not matter whether failure modes are independent.  Unless there is
100% correlation among failures in multiple units (which is never the case)
having more units will increase the overall system reliability.

>From a technical viewpoint, the major drawback to small robots in
>telemetry.  To transmit a signal from Mars, you need a moderately
>large antenna and a bunch of power, if you want reasonable data rates.

What is a "reasonable" data rate?

>I do not have my references in front of me for doing link
>calculations, but from the moon (which is right next door), you need a
>.5 m dish on the moon, putting out 20 W of power to a 10 m dish on the
>earth to get data rates of between 10k-100k bits per second.  The
>problem from mars is much more difficult.  In addition, mars has winds
>which will make antenna pointing difficult, thus degrading data
>transmission.  To overcome these problems, a larger vehicle is needed.
>By larger, I am thinking of a vehicle with a mass of 100-200 kg, as
>opposed to a number of micro rover concepts that have a mass of < 10
>kg.  These micro rovers would be incapable of meaningful scientific
>exploration. 

There you have it, folks: Jerry Roston cannot think of a solution to
the telemetry problem, ergo it is impossible to solve.

First, a lander and orbiter relay system could easily provide high-
bandwidth communications with a small rover.  All of the components
for such a system have been implemented at JPL and tested in field
conditions.

Second, meaningful scientific exploration can be done with *extremely*
low data rates.  A question of current scientific interest is whether
there remain trace amounts of water on Mars.  A flock of microrovers
equipped with water detectors could answer this question by transmitting
back only a single bit of information over the lifetime of the mission.
(There are designs for systems which would not require any transmitting
power at all, but instead rely on corner reflectors which could be
detected by laser beams transmitted from Earth.  This is speculative
technology, but the potential feasability of such a scheme was recently
demonstrated on Galileo.)

Erann Gat
gat@robotics.jpl.nasa.gov

Like I said in the original post, I don't want the "How to Explore
Mars" thread to be about reliability; I have therefore started a new
thread to continue the discussion about reliability.

[I don't speak for the Erebus project; the following are my opinions.]
The Erebus project was not intended to explore reliability, so few
conclusions about reliability should be drawn from it.  That Dante
failed at a single point of failure should not be surprising, given
that a fault-tolerant design was not a goal.  Goals must be limited in
every experimental situation.  The main goal of the Erebus was to test
robotic technologies in a real environment; reliability was not a big
concern since people would be there, and so reliability did not need
to be included as a major goal of the project.

gat@forsight2.jpl.nasa.gov (Erann Gat) writes:
>
>That using multiple small robots increases realiability is not a "naive
>assumption", it is a theoretically and empirically verifiable fact.  It
>does not matter whether failure modes are independent.  Unless there is
>100% correlation among failures in multiple units (which is never the case)
>having more units will increase the overall system reliability.

This is a simplistic statement, which would lead to a hot debate if
made in comp.risks (where it would be phrased "Four engine airplanes
are obviously safer than two engine airplanes, so why aren't there
more of them?").  Two factors this statement ignores are complexity,
and how to maximize reliability while minimizing cost.  Redundancy is
a very important technique for creating reliable systems, but "mere
redundancy" is *not* enough; it is a lot harder than that.  Food for
thought: multiple robot systems already have the very hard problem of
constructive interaction; might this complicate reliability even more?

[begin personal discussion]

gat@forsight2.jpl.nasa.gov (Erann Gat) also writes:
>...
>You know Jerry, I'm really happy when you shoot your mouth off like this
>because it makes me look diplomatic by comparison.
>...
>There you have it, folks: Jerry Roston cannot think of a solution to
>the telemetry problem, ergo it is impossible to solve.

Why are you being so nasty?  If you have a bone to pick with someone,
do it by email; the rest of us are not interested.

[end personal discussion]

- John
--
hagerman@ece.cmu.edu

[As Erann and Gerry cross verbal swords...]

In his post, Erann brought up one important fact: what are you really
trying to accomplish with a mission.  If all you are trying to to is
to determine if there is water at the surface at some point
(literally) on the surface of Mars, the cornet cube concept may be the
appropriate solution.  However, this is a fairly meaningless mission.

Let's backtrak.  IMHO, the reason for exploring the moon and mars is
two-fold: first to gather scientific data for two reasons: pure
science and the economic feasilbility of exploiting resources, for
both in-situ and exportation; and to determine the feasibility of
establishing a permanent manned habitat.   The question being
addressed is whether or not a "micro" rover or a "maxi" rover is more
appropriate to carry out this mission.

The "common wisdom" says this: a "maxi" rover is more capable but the
"micro" rover provides greater reliability.  Is this true? 

Those who have been following this thread know that I am not a
proponent of "micro" rovers.  I will explain some of the difficulties
with these machines, then I will show why they are not a good idea in
reality. 

The biggest drawback with "micro" rovers is their inability to
accomodate components that do not scale in size, for instance
scientific instruments and telemetry systems.  For example, SDIO has
developed a number of imaging sensors.  These are (to the best of my
unclassified world knowledge) the smallest, space qualified imaging
sensors in existance.  For example, a wide field of view camera for
350 grams and about 5 watts of power.  However, when Erann speaks of
"micro rovers" (and Erann, PLEASE correct my impresion if I'm wrong),
he is thinking in terms of machines that mass less than 5 kg.  Thus,
to have only one imaging device consumes 7% of the total available
landed mass!  Now, to do a survey of the type I identified, a suite of
instruments are required.  Okay, I hear the chourus shouting, send
many "micro" rovers, each with one instrument in to the same area.
Bzzt!  How do you register the data returned by one rover with the
data returned by another?  This is a very difficult problem. 

Other difficulties with "micro" rovers is the small surface area
presented (needed for solar power - an assumption, RTG type sources
could be used, but at 1 thermal watt per 30 grams and a 5% conversion
efficiency, this would consume too much landed mass to produce any
reasonable amount of power.  Of course, you could store the energy in a
battery/capacitor to get higher power, but this again consumes
mass...)  Of course, the biggest problem with "micro" rovers is this:
Because of the lack of scaling, the mass fraction of a micro rover
that can be used for scientific purposes will be quite limited.  Let
us say, giving them the benefit of the doubt, that 25% can be thus
allocated.  Having many micro rovers does not change this mass
fraction.  We (FRC) are currently working on the Ambler's successor, a
combined lander/rover that can deliver approximately 50% of the landed
mass as scientific payload.  

Now, let me lastly address one of Erann's comments, specifically,
using a relay satellite.  This solution has two difficulties: first,
the satellite will not be able to continuously communicate between any
given "micro" rover and the earth.  This necessitates, therefor, the
ability of "micro" rover to store and forward collected data, or to
sit idle.  This will have the overall effect of decreasing the amount
of data returned to the earth.  The second problem is that the
satellite is free: it costs additonal dollars and consumes deliverable
mass on launch vehicle, further decreasing the deliverable scientific
mass to the planetary surface.  

As a final note about the corner cube concept.  Let us assume that a
given launch vehicle (say a Delta II) can deliver 500 kg to the
Martain surface.  (It is capable of launching a vehicle that can soft
land 450 kg on the lunar surface, so this seemed like a reasonable
estimate.)  The "micro" rovers to be delivered each have a mass of 1
grams and can determine if there is water underneath them.  Due to
their low mass, these rovers are simply seeded over the surface.  So,
we have 500,000 rovers being delivered.  Now, assuming they all
function and all report a positive finding, this is a mere 500,000
bits of information, the same as only two video images.  So, this may
seem like a neat idea, but in the analysis, it doesn't seem too
worthwhile. 

Finally, lets deal with reality.  There are only a fixed number of
launch vehicles available in the world.  Since this represents the
single largest cost item in a mission, you want to maximize
utilization of this resource, both its mass and volume.  LV payload
fairing are basically cylindrical in shape.  To maximize utilization
of this volume, the "best" shape is a cylinder that fits snuggly
within this volume.  By using multiple, smaller objects, volume
utilization will fall far short of optimal.  Also, due to launch
loads, etc, the multiple smaller payloads will have to be fixed in
place by some sort of bracing structure, yet again additonal mass that
reduces deliverable scientific payload to the surface.

To directly address the question of big versus small, let me make some
comments about the Ambler.  The Ambler's critiques frequently point to
its large size as being its most serious drawback, but this is not the
case.  The Ambler, as designed, not as built, could have been
accomodated by a Titan IV launch vehicle.  But then again, this was
not one of hte driving design issues at the time the Ambler was built.
Its most serious drawback was its immense complexity, however, this
complexity allowed the Ambler an unprecedented range of mobility, a
range that no "micro" rover could ever hope to mimick.  

Now, in conclusion (finally!), am I saying that making robots the
size of the Ambler is appropriate?  Probably not, though it depends on
the goal of the specific mission.  By the same token, however, I will
say that there appear to be no missions for which a "micro"rover is
appropriate.  It seems to me that the "best" sized rover is one that
will completely consume the available payload of a cost-efficient
launch vehicle, and use this "large" landed mass for carying a
meaningful suite of scientific instruments, along with the capability
to deliver these instruments to a wide variety of locations with a
high degree of competency.

--
Gerry Roston (gerry@cmu.edu)    | We hold these truths to be self-evident, that
Field Robotics Center,          | all men are created equal, that they are
Carnegie Mellon University      | endowed by their Creator with certain
Pittsburgh, PA, 15213           | unalienable Rights, that among these are
(412) 268-3856                  | Life, Liberty, and the pursuit of Happiness.
                                | That to secure these rights, Governments are
The opinions expressed are mine | instituted among Men, deriving their just
and do not reflect the official | powers from the consent of the governed.
position of CMU, FRC, RedZone,  | That whenever any Form of Government becomes
or any other organization.      | destructive of these ends, it is the Right of
                                | the People to alter or to abolish it...
                                | Thomas Jefferson

In article <HAGERMAN.93Jan11230044@rx7.ece.cmu.edu> hagerman@ece.cmu.edu (John Hagerman) writes:
>[I don't speak for the Erebus project; the following are my opinions.]
>The Erebus project was not intended to explore reliability, so few
>conclusions about reliability should be drawn from it.  That Dante
>failed at a single point of failure should not be surprising, given
>that a fault-tolerant design was not a goal.  Goals must be limited in
>every experimental situation.  The main goal of the Erebus was to test
>robotic technologies in a real environment; reliability was not a big
>concern since people would be there, and so reliability did not need
>to be included as a major goal of the project.

It's fair enough to say that fault tolerant design was not a goal of the
design, but surely the system should be fault tolerant enough to
acomplish the design goals.  By saying "people would be there" is surely
a catch-all phrase - if system X did not work, then it's OK since people
would be there to replace it.  This could cover all possibilities - it's
OK that the legs didn't work, people were there to carry it....

>gat@forsight2.jpl.nasa.gov (Erann Gat) writes:
>>
>>That using multiple small robots increases realiability is not a "naive
>>assumption", it is a theoretically and empirically verifiable fact.  It
>>does not matter whether failure modes are independent.  Unless there is
>>100% correlation among failures in multiple units (which is never the case)
>>having more units will increase the overall system reliability.
>
>This is a simplistic statement, which would lead to a hot debate if
>made in comp.risks (where it would be phrased "Four engine airplanes
>are obviously safer than two engine airplanes, so why aren't there
>more of them?").  Two factors this statement ignores are complexity,
>and how to maximize reliability while minimizing cost.  Redundancy is
>a very important technique for creating reliable systems, but "mere
>redundancy" is *not* enough; it is a lot harder than that.  Food for
>thought: multiple robot systems already have the very hard problem of
>constructive interaction; might this complicate reliability even more?

Granted, constructive interaction is a problem right now.  The answer is
to have *no* interaction.  Send X completely independant robots down the
volcano, recoding from each.  It is unlikely that all of them would
encounter the same fault/environmental problem.  When you compare the
unit costs per robot to the over all design costs, I bet that making a
few robots once the design were finalised would make very little change
to the overall costs of the project.

Bill

wsmart@uk.ac.dund.mcs