geez
[Top] [All Lists]

Re: Confusing G-Cube desktop "sanity check"

To: rasmussend@mindspring.com
Subject: Re: Confusing G-Cube desktop "sanity check"
From: Gary Browning <garyb@robotics.Eng.Sun.COM>
Date: Tue, 16 Apr 2002 17:52:39 -0700 (PDT)
Dick,

Dick Rasmussen <rasmussend@mindspring.com> wrote:
> At 07:40 PM 4/15/02, Gary Browning wrote:
> 
> >But the map was not what I expected.  I had expected a pattern of an upside
> >down 'T' like the pattern I had moved the G-Cube through.  What I got was a
> >line that started at the bottom and went vertically up.  There were no left 
>or
> >right, just vertical !!  The L/R strip chart shows 0.2G left at one point,
> >but map shows no left movement.  It does show vertical movement though.
> 
> Gary,
> 
> I'm not clear on what you actually did ...

It was pointed out to me that the lateral acceleration values are treated
differently than the forward/reverse acceleration values when generating the 
map.
The test I did was not a real world one like I expected because the "car" had
lateral acceleration with no velocity (standing still).  It is pretty hard for a
car to go sideways unless it has some forward/reverse velocity.
The lateral acceleration values are used with calculated velocity to
determine the radius of the curve.  No velocity, no radius.  Doh!

> ... but you might try this:
> 
> Use your normal settings including autostart.
> 
> Tip cube first on say the back side (simulates a start), then the left 
> side, then the right side, then the front. (just like during calibration) 
> Repeat a few times just to be sure you have started the recording and keep 
> it going long enough to get a minimum run.
> 
> When the cube is actually on a side, front, or back the G reading should be 
> 1 (if I recall) in the relevant direction. The lateral g's would alternate 
> between 1 g left, zero, and 1 g right. Accel would alternate between 1 g 
> accel, zero, and 1 g braking. Basically lots of "steps" if you are smooth 
> but relatively quick with the changes.
> 
> It has "been awhile" so I don't guarantee that this is correct:-)

Regarding a better checkout procedure, something akin to what you have suggested
should work.  I'll pass along what was suggested to me -- place the bottom 
against
a wall and the rotate it to test each direction (front, right, back, left).  The
lateral and for/aft strip charts should be checked and the total g's strip chart
will always be 1g.  I'll have to give this a try.

-- Gary

///  Archives at http://www.team.net/archive
///  Check http://www.team.net/mailman/listinfo for other lists of interest.


<Prev in Thread] Current Thread [Next in Thread>