In-flight auto-repair spaceprobes via. nanobots
Henry Spencer
Javan.Stephen <Javan....@btinternet.com> wrote:
>How far are we technically from being able to build spaceprobes/craft that
>once in deep space missions, can recieve simple commands from operators
>mobilise onboard nanobots (or "micro" bots) that can conduct simple physical
>operations such as bending a panel or removing/refitting/cleaning a lens.
We're a long way from being able to do that on Earth, never mind on a
spacecraft.
--
When failure is not an option, success | Henry Spencer he...@spsystems.net
can get expensive. -- Peter Stibrany | (aka he...@zoo.toronto.edu)
Mike Rhino
replace either the entire set or an entire circuit board or speaker or
something. In some ways we are moving away from being able to fix things.
It might be possible to launch 3 spacecraft that later condensed down to 2
spacecraft and a container of spare parts. Some work would have to be done
on designing a spacecraft that could be repaired. I'm not aware of any such
research. There is some risk that the same part could wear out in all three
spacecraft after 10 years.
I don't think that nanobots are the best way to go. A larger robot with the
strength of a human auto-mechanic would be better. Nanobots could be useful
for diagnosing problems. This might be a useful feature for a space colony
where people's lives are on the line. A robot might be better at fixing
things than a human wearing a space suit.
"Javan.Stephen" <Javan....@btinternet.com> wrote in message
news:972sr0$13h$1...@uranium.btinternet.com...
> How far are we technically from being able to build spaceprobes/craft that
> once in deep space missions, can recieve simple commands from operators
> mobilise onboard nanobots (or "micro" bots) that can conduct simple
physical
> operations such as bending a panel or removing/refitting/cleaning a lens.
If
> the operation is performed command-by-command (in semi-VR) surely this
> operations aren't that difficult since it doesnt' require much AI of the
> bot??? I
rk
>
> In the old days, TV repairmen tracked down which tube was bad. Now people
> replace either the entire set or an entire circuit board or speaker or
> something. In some ways we are moving away from being able to fix things.
In the old days, components were very expensive. Now they are quite
cheap. Labor is expensive. Whether to diagnose and repair versus
replace is an economic decision. Hint, from the local computer store:
Intel Celeron MMX 600MHz (S370-128K) & Motherboard: $149.00
Let's say the motherboard fails. Cost diagnosis + parts cost + repair
costs versus replacement costs. And note that a lot of the packages
used on these chips do not just plug in and out like a tube, testable on
a simple standard machine in the neighborhood hardware store. These
test machines are *expensive.*
> It might be possible to launch 3 spacecraft that later condensed down to 2
> spacecraft and a container of spare parts.
It would be cheaper to launch 5 spacecraft without the ability to
rendezvous and fix each other.
Hints:
1. Look at component reliability for modern equipment.
2. Look at the success record of space probes and see if
any of this is worthwhile.
3. Look at all of the failures on space probes and see how
many would be saved by nanobots.
> Some work would have to be done
> on designing a spacecraft that could be repaired. I'm not aware of any such
> research. There is some risk that the same part could wear out in all three
> spacecraft after 10 years.
Correct. But wearout is typically not a major factor. Random failures
are. The parts are being operated (or should be!) in the flat part of
the bathtub reliability curve. Heck, it's a good exercise [one I
recently did] to compare the failures rate in FITS for modern commercial
electronics versus hi-rel electronics from the '80s.
> I don't think that nanobots are the best way to go. A larger robot with the
> strength of a human auto-mechanic would be better. Nanobots could be useful
> for diagnosing problems. This might be a useful feature for a space colony
> where people's lives are on the line. A robot might be better at fixing
> things than a human wearing a space suit.
For the near, medium, and perhaps long term, simple redundancy and
fault-tolerance techniques for spaceprobes will be more cost-effective
than nanobots. These techniques are straightforward and readily
implemented.
----------------------------------------------------------------------
rk "There will be prayer in schools
stellar engineering, ltd. as long as schools give tests"
stel...@erols.com.NOSPAM -- can't remember who said it - 2000
Hi-Rel Digital Systems Design
Peter Knutsen
"Javan.Stephen" wrote:
>
> How far are we technically from being able to build spaceprobes/craft that
> once in deep space missions, can recieve simple commands from operators
> mobilise onboard nanobots (or "micro" bots) that can conduct simple physical
> operations such as bending a panel or removing/refitting/cleaning a lens. If
> the operation is performed command-by-command (in semi-VR) surely this
> operations aren't that difficult since it doesnt' require much AI of the
> bot??? I
I think the biggest problem is that we don't have microbots yet, let
alone nanobots.
Also, you're talking about large-scale repair and I find it very hard
to see how microbots or nanobots can exert the force necessary to
"bend panels" and other such things that happens on the millimetre-
scale if not larger.
--
Peter Knutsen
Brett Buck
Javan....@btinternet.com wrote on 2/22/01 12:34 AM:
> How far are we technically from being able to build spaceprobes/craft that
> once in deep space missions, can recieve simple commands from operators
> mobilise onboard nanobots (or "micro" bots) that can conduct simple physical
> operations such as bending a panel or removing/refitting/cleaning a lens. If
> the operation is performed command-by-command (in semi-VR) surely this
> operations aren't that difficult since it doesnt' require much AI of the
> bot??? I
>
We are VERY far away from that. Conceptually, you just increase the
complexity of the system and make it less reliable, the more parts you add.
Spacecraft with "autonomy systems" tend to be much more prone to certain
kinds of errors, like S/W and database/script errors, than non-autonomous
system. Deep-space mission need some degree of this, due to the
impossibility of live or near-live interaction, but the ones that have
worked the best have limited this to relatively simple,
just-good-enough-to-insure-power-and-contact-with-the-ground systems. The
NEAR fly-by anomaly and the Lewis loss are good examples. The Lewis problem
was a particularly painful example, as the autonomy system was used to
justify cheaping out on the ops support. Doesn't seem so cheap now.
After seeing a number of autonomy systems, the rule seems to be that the
more autonomy software you have, the more people you need to outthink it,
and the sharper they have to be. And this is with the relatively simple
state of the art. Your hypothetical system would be several orders of
magnitude further down the road, and will require some breakthroughs in AI
that don't appear to be imminent.
It's just my opinion (based on what I've seen happen with autonomy
systems), that given the current level of technology you are much better off
making the spacecraft simple and reliable enough to do the mission so it
doesn't frequently break to begin with, rather than to try to push the
parts reliability and then depend on autonomous actions to fix it.
Brett
Brett
Gary Weiner
Peter Knutsen wrote:
>
> "Javan.Stephen" wrote:
> >
> > How far are we technically from being able to build spaceprobes/craft that
> > once in deep space missions, can recieve simple commands from operators
> > mobilise onboard nanobots (or "micro" bots) that can conduct simple physical
> > operations such as bending a panel or removing/refitting/cleaning a lens. If
> > the operation is performed command-by-command (in semi-VR) surely this
> > operations aren't that difficult since it doesnt' require much AI of the
> > bot??? I
>
> I think the biggest problem is that we don't have microbots yet, let
> alone nanobots.
Not completely true
http://www.sandia.gov/media/NewsRel/NR2001/minirobot.htm
>
> Also, you're talking about large-scale repair and I find it very hard
> to see how microbots or nanobots can exert the force necessary to
> "bend panels" and other such things that happens on the millimetre-
> scale if not larger.
I agree. What repairs could such small robots perform?
--
Gary J. Weiner \ "When you can balance a tack hammer on
webm...@hatrack.net \ your head, you will head off your foes
http://www.hatrack.net \ with a balanced attack!"
"Hang Your Web With Us!"\ -The Sphinx "Mystery Men"
Jonathan Thornburg
Brett Buck <buc...@pacbell.net> wrote:
>Spacecraft with "autonomy systems" tend to be much more prone to certain
>kinds of errors, like S/W and database/script errors, than non-autonomous
For an excellent review of the difficulties, and state of the art
in autonomous spacecraft as of about 5 years ago, see
> ## http://www-aig.jpl.nasa.gov/public/home/gat/gp.html
>
> News From the Trenches: An Overview of Unmanned Spacecraft for AI
> Researchers
>
> Erann Gat
> Jet Propulsion Laboratory
> California Institute of Technology
> 4800 Oak Grove Drive
> Pasadena, CA 91109
> g...@jpl.nasa.gov
>
> Abstract
>
> This paper is a brief description of certain aspects of the problem of
> controlling autonomous spacecraft that are relevant for AI
> researchers. It is argued that this is a worthwhile domain for
> implementing a wide variety of AI theories.
>
> This paper was written for the 1996 AAAI Spring Symposium on Planning
> with Incomplete Information for Robot Problems.
--
-- Jonathan Thornburg <jth...@thp.univie.ac.at>
http://www.thp.univie.ac.at/~jthorn/home.html
Universitaet Wien (Vienna, Austria) / Institut fuer Theoretische Physik
Q: Only 6 countries have the death penalty for children. Which are they?
A: Congo, Iran, Nigeria, (Pakistan[*]), Saudi Arabia, United States, Yemen
[*] Pakistan reportedly ended it in July 2000. -- Amnesty International
http://www.web.amnesty.org/ai.nsf/index/AMR511392000
Peter Knutsen
Gary Weiner wrote:
>
> Peter Knutsen wrote:
> > I think the biggest problem is that we don't have microbots yet, let
> > alone nanobots.
>
> Not completely true
>
> http://www.sandia.gov/media/NewsRel/NR2001/minirobot.htm
That's a minirobot, not a microbot. "micro" denotes a scale
of a millionth of a meter, "nano" denotes a scale of a
billionth of a meter. Your minirobot is only on the "milli"
scale.
> --
> Gary J. Weiner
--
Peter Knutsen