Newsgroups: comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!ix.netcom.com!netcom.com!kovsky
From: kovsky@netcom.com (Bob Kovsky)
Subject: Re: Robot autonomy (was Is the mind/brain deterministic?)
Message-ID: <kovskyCzGyMx.ELv@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <kovskyCzF8D4.Bxv@netcom.com> <3aghdc$q3s@cantaloupe.srv.cs.cmu.edu>
Date: Fri, 18 Nov 1994 15:20:57 GMT
Lines: 53

In a discussion about whether robots could navigate in a wilderness, I 
wrote: 


>> My understanding is that humans guided the navigation of these
>> [planetary exploration] devices, by remote control.  My understanding
>> is that Prof. Moravec's robots are intended to be autonomous and
>> self-navigating.  I will be very interested if my understanding is in
>> error.
> 

Prof. Moravec responded:

>CMU's Dante II (a prototype for a planetary explorer) that descended
>into Alaska's Mt. Spur a few months ago, uses a scanning laser
>rangefinder to model the terrain around it, and plans its paths and
>foot placements on that basis.  It is capable of fully autonomous
>travel, and has walked distances of hundreds of meters on its own in
>slag heaps and down and up steep slopes in its testing phases around
>Pittsburgh.  For the Mt. Spur excursion, and probably in planetary
>use, it is prudent to let a remote committee of experts help it in its
>decision making, by pointing out places to go, or even to step.  But
>the robot can go arbitrarily far on its own decisions.  On its way
>back out after exploring the crater floor for a few days, the robot
>was walking autonomously, retracing its path in, when the icy, muddy,
>bubbly ground at one point gave way, having thawed since the trip
>down.  After the robot toppled, it was still scanning and searching
>for a sequence of moves that would to regain its horizontal
>orientation, unfortunately such a solution did not exist.

(References omitted)

	A superb achievement, which has not been widely enought reported 
to come to my attention.  Your Alaska trip must have been a wonderfully 
memorable experience.  

	My skepticism about the possiblity of self-navigating robotic
travel was based on my understanding of current technology, not on
principle.  My views do not exclude the possibility.  I anticipated that
the amount of computation would be beyond the reach of programmers and
miniatarizers:  apparently not.  A regular polygon with 100 million sides
can approach a circle pretty darn close.  The Singh article you cite (from
1991) mentioned a robot (or vehicle) that used "information from inertial
and satellite instruments to navigate."  Do Dante and Dante II also use
information from satellites?  Also:  if you have achieved success in
wilderness, what is different about the indoor robot to be used for floor
cleaning and delivery that makes it a more difficult problem? 

-- 

*   *    *    *    *    *    *    *    *    *    *    *    *    *    *    *   * 
    Bob Kovsky          |  A Natural Science of Freedom 
    kovsky@netcom.com   |  Materials available by anonymous ftp
                        |  At ftp.netcom.com/pub/freeedom
*   *    *    *    *    *    *    *    *    *    *    *    *    *    *    *   * 
