geez
[Top] [All Lists]

Re: How does GEEZ calculate usage?

To: Todd Green <tag@cs.utah.edu>
Subject: Re: How does GEEZ calculate usage?
From: Brian M Kennedy <kennedy@i2.com>
Date: Thu, 28 Oct 1999 13:37:52 -0500
At 12:01 PM 10/27/99 , Todd Green wrote:
>Mark Sipe sent some private email which got me thinking.  GEEZ
>appears to determine usage via:
>
>   usage = max(current lat, current accel, current lat trans) / 
>           max(peak lat, peak accel, peak lat trans)

Hmmm.  That would seem flawed.  You could be right on the edge of the
friction circle, at a 45 degree point, and get a usage in the 80's,
assuming you normalized.  If you didn't, it would be even more flawed.

Ignoring the transition component, you'd want something like:

        g_usage = sqrt((cur_lat/peak_lat)^2 + (cur_accel/peak_accel)^2)

where you use a different peak_accel depending upon the sign of cur_accel.

Perhaps you would 'max' that g_usage with transition rate:

        usage = max(g_usage, cur_trans/peak_trans)

Not sure the best way to incorporate transition rate, but don't you need
one for lateral and one for acceleration?  And perhaps even two for 
accel, one for throttle->brake and one for brake->throttle, taking into
account the slower "spool up" of motors vs. brakes???

Well, I am sure Byron will dazzle us well beyond the above.  ;^)

Until then, that's just food for thought,

Brian


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