Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!samsung!think.com!news.arc.nasa.gov!ncar!gatech!prism!jm59
From: jm59@prism.gatech.EDU (MILLS,JOHN M.)
Newsgroups: comp.robotics
Subject: Re: Small Hovering Hobby Robots
Message-ID: <31714@hydra.gatech.EDU>
Date: 24 Jun 91 13:22:23 GMT
References: <25523@well.sf.ca.us> <1991Jun20.155520.12703@sagpd1> <1991Jun23.095642.872@jarvis.csri.toronto.edu>
Organization: Georgia Institute of Technology
Lines: 54

In article <1991Jun23.095642.872@jarvis.csri.toronto.edu> mccool@dgp.toronto.edu (Michael McCool) writes:
>I've been following this line for a while, and thought I would put in my
>two cent's worth of ridiculous suggestions.
>

I have a one of those [suggestions] myself:
For a platform hovering in the ground-effect range, you should be able to
provide passive stabilization:
  (1) Break up the plenum under the platform into sections around the
edge, or implement it as "pads" at several points around the edge, kind of
like feet, and _isolate_ the air supply to the various pads either:
    (a) use separate air supplies, or (better)
    (b) bleed the air to each plenum through a separate restrictor from
        a central fan.
As the platform tilts, the exit areas around the low pads are reduced, and
the areas around the high-side pads increase.  The idea is to make this
inherently reduce the pressure at the high corner(s), so they won't lift
further, and increase the pressure under the low corner(s) so they won't
sink further. EE types can sketch this as a resistor network.  The 
restrictors between the plena _are_ needed.  Trimming the pressure in
the various pads should allow the platform to float level, even its
CG is off-center w.r.t the pads. (Pad _area_ may also be adjusted to
balance an off-center load.)

This should make the system inherently stable, so little or no control will
be needed for leveling.  Always easier to try to use the physics of the
situation than to fight it, no?

Now _my_ ridiculous question:
How could one keep the thing from simply skidding off downhill?  I don't
think this depends primarily on the platform's horizontal attitude,
though the "high corner" pads _could_ in principle add a [small] directional
thrust component.  Anyone have a simple way around this?

>For forward motion, just have some way of shifting the center of gravity 
>off-center, which will tilt the robot forward (actually, you'd have to 
>deal with the gyro effect, of course... have fun).  So, you don't need
>a thruster.  If you had the weight-shifting under computer control you
>could probably do dynamic balancing and not need an ungainly T shape.
>You are going to need some pretty sophisticated navigation anyways for 
>a powered hovering vehicle.  The robot would end up looking like a flying
>saucer.   

I don't think this works very well for "hovercraft" type gadgets --
Thrust vectoring sounds better to me, except maybe for a helicopter.
I may be wrong, since I haven't built one. That's why I asked.

Thanks for any comments.

-- 
MILLS,JOHN M.
Georgia Institute of Technology, Atlanta Georgia, 30332
uucp:	  ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!jm59
Internet: jm59@prism.gatech.edu