NASA and robotic research

http://www.stnews.org/News-528.htm
At last there is news that NASA is doing some serious robotics
research. I feel though that NASA is thinking in a far too
anthropomorphic way. If you are talking about LEO telepresence I
suppose that anthropomorphism is to a degree necessary. However where
the return light journey is a matter of minutes or hours I would
advocate something like a hub with manipulators and sensors attached. I
note that robot - robot communication is envisaged as is "speech and
gestures". Why not simply have the normal protocols of computer -
computer communication - Ethernet , WiFi, Bluetooth, USB or Firewire?
gestures/speech. If telepresent one communicates by pressure on the
suit If Mars (telepresence impossible) simply send something
resembling an E-mail.

BTW If the normal computer - computer protocols are used we can perform
a task using UNIX and PVM (Parallel Virtual Machine).

I note too that decision making and geometrical understanding is going
to be made a priority. I have said before that the basic step for a Von
Neumann machine was understanding CAD/CAM. The article did not mention
building and repairing robots, but this is implicit.

wrote:
http://www.stnews.org/News-528.htm
At last there is news that NASA is doing some serious robotics
research. I feel though that NASA is thinking in a far too
anthropomorphic way. If you are talking about LEO telepresence I
suppose that anthropomorphism is to a degree necessary. However where
the return light journey is a matter of minutes or hours I would
advocate something like a hub with manipulators and sensors attached. I
note that robot - robot communication is envisaged as is "speech and
gestures". Why not simply have the normal protocols of computer -
computer communication - Ethernet , WiFi, Bluetooth, USB or Firewire?
gestures/speech. If telepresent one communicates by pressure on the
suit If Mars (telepresence impossible) simply send something
resembling an E-mail.

BTW If the normal computer - computer protocols are used we can perform
a task using UNIX and PVM (Parallel Virtual Machine).

I note too that decision making and geometrical understanding is going
to be made a priority. I have said before that the basic step for a Von
Neumann machine was understanding CAD/CAM. The article did not mention
building and repairing robots, but this is implicit.

Well that's why it's anthromorphuc.
CAD/CAM in 0-G is not a Von Neumann machine.

It is because everything is defined in terms of CAD/CAM. It is
inconceivable that a robot would not in itself be so defined. This
being the case all the processes needed to construct such a robot,
including the basic extraction processes would be thus defined.

No it is not a VN machine, but after CAD/CAM the momentum towards one
is irresistable.

BTW - We should be speaking of a VN swarm. If we send 2 robots + spares
to Mars one can repair the other.

wrote:
It is because everything is defined in terms of CAD/CAM. It is
inconceivable that a robot would not in itself be so defined. This
being the case all the processes needed to construct such a robot,
including the basic extraction processes would be thus defined.

No it is not a VN machine, but after CAD/CAM the momentum towards one
is irresistable.

We know that, just like the sun is irrestibale.
But the engineering problems are the same as
submarine problems, and concern the shape of
lungs, not brains.




BTW - We should be speaking of a VN swarm. If we send 2 robots + spares
to Mars one can repair the other.

We know that, just like the sun is irrestibale.
But the engineering problems are the same as
submarine problems, and concern the shape of
lungs, not brains.

I find this hard to follow. A robot is electrically powered, it needs
no lungs. The only reason why you want anthropomorphism is in LEO where
you can be telepresent, get into a suit and become the robot. This is
impossible at distances greater than about 0.2 sec (60,000km).

For telepresence of couse no understanding, other than transfer
functions, is needed as a human is providing all the intelligence.

The basic reasoning which will lead to VN capabilities is the ability
to understand the CAD/CAM environment, assemble assemblies from parts
and sub assemblies, and carry out instructions to build something,
defined in CAD terms. Understanding must, of course, be such as to
carry out these functions unaided.

The nature of the operating environment is to some extent irrelevant,
except to say that it must be understood, and the robot must have the
basic capabilities within whatever environment it is operating.

In article .com,
wrote:
http://www.stnews.org/News-528.htm
At last there is news that NASA is doing some serious robotics research...

Uh, notice that said "news" is a year old. And it's probably obsolete,
because NASA's latest budget is drastically cutting many R&D areas which
are seen as not immediately relevant to the return to the Moon. As Jeff
Greason put it at Space Access: "NASA is well and truly out of the R&D
business; they're going to focus on their core competencies."
--
spsystems.net is temporarily off the air; | Henry Spencer
mail to henry at zoo.utoronto.ca instead. |

Henry Spencer wrote:

As Jeff
Greason put it at Space Access: "NASA is well and truly out of the R&D
business; they're going to focus on their core competencies."
--
I wonder what those are? Any thoughts?

Henry Spencer wrote:
In article .com,
wrote:
http://www.stnews.org/News-528.htm
At last there is news that NASA is doing some serious robotics research...

Uh, notice that said "news" is a year old. And it's probably obsolete,
because NASA's latest budget is drastically cutting many R&D areas which
are seen as not immediately relevant to the return to the Moon. As Jeff
Greason put it at Space Access: "NASA is well and truly out of the R&D
business; they're going to focus on their core competencies."
--
spsystems.net is temporarily off the air; | Henry Spencer
mail to henry at zoo.utoronto.ca instead. |

To me it is obvious that robotics are the key to space. The ISS (and
the Shuttle) which is costing all the money is going precisely nowhere.
A golf shot - Big big deal!

A determined effort, spending a lot lot less than on the above items,
will somewhere down the line create a VN machine. I suppose being
British I should welcome this, but if the US does not do it other
countries will. A lead in robotics will represent a lead in space in
general, such is its importance.

If what you say is true, the US has lost its lead in space.

Alex Terrell wrote:

Greason put it at Space Access: "NASA is well and truly out of the R&D
business; they're going to focus on their core competencies."

I wonder what those are? Any thoughts?

Consuming federal funds.

Paul

wrote:
We know that, just like the sun is irrestibale.
But the engineering problems are the same as
submarine problems, and concern the shape of
lungs, not brains.

I find this hard to follow. A robot is electrically powered, it needs
no lungs. The only reason why you want anthropomorphism is in LEO where
you can be telepresent, get into a suit and become the robot. This is
impossible at distances greater than about 0.2 sec (60,000km).

For telepresence of couse no understanding, other than transfer
functions, is needed as a human is providing all the intelligence.

The basic reasoning which will lead to VN capabilities is the ability
to understand the CAD/CAM environment, assemble assemblies from parts
and sub assemblies, and carry out instructions to build something,
defined in CAD terms. Understanding must, of course, be such as to
carry out these functions unaided.

The nature of the operating environment is to some extent irrelevant,
except to say that it must be understood, and the robot must have the
basic capabilities within whatever environment it is operating.

Full conciousnes and "understanding unaided" as you put it, is not
necessary. Hybrid approach somewhere between full telepresence and full
autonomy is entirely possible. You can leave difficult and higher-level
decisions still to humans while providing sufficient autonomy to carry
out simple operations. Also sensory processing and congitive tasks that
are still difficult for robots can be delegated to human operators ( as
in, choosing the best travelling route based on visual data, where to
dig, what to do in unexpected situations etc )
This is the case with MER for example, it has sufficient autonomy to
get along with its tasks for a short time, but is still guided by
humans. Its just that two MERs have a pack of highly paid scientists
guiding it, for real work you would want one operator guiding and
assisting a pack of robots.

-kert