🚀 go-pugleaf

In article <8528@jpl-devvax.JPL.NASA.GOV>, kandt@ai-jupiter.JPL.NASA.GOV (Kirk Kandt) writes:
> A long time ago when I worked in computer vision at Hughes Research Labs
> we had a project funded by Army/DARPA for autonomous vehicles.  This was
> before the Strategic Computing Project.  The biggest technological
> problem that we had was to prevent the vehicle from falling into bomb
> craters, and if it did to get it out.  In real-world environments a
> vision system will fail (on occasion) to detect such hazards.  In such a
> case, the vehicle must understand what occured and rely on mechanical
> means to get out.  We found a millipede-like object (on wheels) that
> could climb loading docks.  This platform was large (about 12 feet)
> which gave it the ability to get out.  So, if a mini-rover happens to
> fall into a moon crater, for example, how does it get out? 

The basic idea here is that if a mini-rover (or micro-rover) falls in
a crater or whatever: (1) it will be small enough that it will be less
susceptible to serious damage than a larger "moon stomper", and (2) it
will try to get out of the crater more or less like a roach or other
insect would try to get out of, say, a toilet bowl here.  Granted, it
would probably need a bit more intelligence than the current crop of
MIT mini-rovers, but it won't require the computing power of older
rovers -- that are supposed to understand a good bit about their
surroundings.  If the simple-minded approach doesn't work, the
mini-rover will (in theory) be cheap enough that many could be
send on an exploratory mission, and it could be left behind without
making a serious dent in the mission budget -- or seriously endangering
the overall mission goals.  
-- 
What?  Me speak for my  | William H. Ganoe      bill@tucson.sie.arizona.edu
employer?               |  Systems & Industrial Engr. Dept, Univ. of Arizona
                        |  Tucson, AZ 85721; USA