From: Lev Bishop (lev.bishop_at_yale.edu)
Date: Fri Jul 12 2002 - 02:43:58 CEST
Return-Path: <owner-cavexml-outgoing_at_ethz.ch> Delivered-To: cavexml-archive_at_cartography.ch Received: from localhost (localhost [127.0.0.1]) by karmail.ethz.ch (Postfix on SuSE eMail Server 2.0) with ESMTP id 6984714AB1 for <cavexml-outgoing_at_ethz.ch>; Fri, 12 Jul 2002 02:54:05 +0200 (CEST) Received: by karmail.ethz.ch (Postfix on SuSE eMail Server 2.0, from userid 28) id 41BE414A6E; Fri, 12 Jul 2002 02:54:03 +0200 (CEST) Delivered-To: cavexml-loopcheck_at_ethz.ch Received: from localhost (localhost [127.0.0.1]) by karmail.ethz.ch (Postfix on SuSE eMail Server 2.0) with ESMTP id 3325014AB1 for <cavexml-loopcheck_at_ethz.ch>; Fri, 12 Jul 2002 02:54:02 +0200 (CEST) Received: by karmail.ethz.ch (Postfix on SuSE eMail Server 2.0, from userid 96) id 07BE28A96; Fri, 12 Jul 2002 02:53:59 +0200 (CEST) Delivered-To: cavexml_at_cartography.ch Received: from localhost (localhost [127.0.0.1]) by karmail.ethz.ch (Postfix on SuSE eMail Server 2.0) with ESMTP id BA54214AB1 for <cavexml_at_cartography.ch>; Fri, 12 Jul 2002 02:53:59 +0200 (CEST) Received: from pantheon-po02.its.yale.edu (pantheon-po02.its.yale.edu [130.132.50.17]) by karmail.ethz.ch (Postfix on SuSE eMail Server 2.0) with ESMTP id F1C197D84 for <cavexml_at_cartography.ch>; Fri, 12 Jul 2002 02:53:56 +0200 (CEST) Received: from ares.its.yale.edu (ares.its.yale.edu [130.132.52.16]) by pantheon-po02.its.yale.edu (8.11.6/8.11.6) with ESMTP id g6C0hux04114 for <cavexml_at_cartography.ch>; Thu, 11 Jul 2002 20:43:56 -0400 (EDT) Received: from localhost (lsb32_at_localhost) by ares.its.yale.edu (8.11.6/8.11.6) with ESMTP id g6C0hw529361 for <cavexml_at_cartography.ch>; Thu, 11 Jul 2002 20:43:58 -0400 X-Authentication-Warning: ares.its.yale.edu: lsb32 owned process doing -bs Date: Thu, 11 Jul 2002 20:43:58 -0400 (EDT) From: Lev Bishop <lev.bishop_at_yale.edu> To: cavexml_at_cartography.ch Subject: Re: comments on the data model In-Reply-To: <3D2DD9A9.8080201_at_aic.nrl.navy.mil> Message-ID: <Pine.LNX.4.44.0207112023420.25882-100000@ares.its.yale.edu> Content-Type: TEXT/PLAIN; charset=US-ASCII X-YaleITSMailFilter: Version 1.0c (attachment(s) not renamed) X-Loop: cavexml Sender: owner-cavexml_at_karmail.ethz.ch Precedence: bulk Reply-To: cavexml_at_cartography.ch X-Virus-Scanned: by AMaViS perl-11
On Thu, 11 Jul 2002, Ralph Hartley wrote:
> Richard Knapp wrote:
>
> > When COMPASS does loop closure
> > ("corrected data"), it creates a copy of the DAT file with new
> > distances, azimuths, and inclinations (and sometimes survey ties).
>
> Does it?
>
> If true, this would be rather bizare, reducing the data to get best
> fitting x,y,z coordinates, and then converting them back into "corrected
> survey shots".
Actually its not all that bizarre, really. If you were to solve the "full"
survey problem, as a nonlinear maximum-likelihood estimation problem, then
the only way is to (iteratively) solve for the corrected instrument
readings. The xyz coords drop out at the end as a bonus, andwith some ways
of programming it, you may never even calculate them during loop closure.
That said, all survey software that I'm aware of doesn't actually solve
the full survey problem but rather linearizes it by assuming that each
shot produces a vector with a 3d gaussian distribution as its error. At
this point linear least squares can be used to solve the problem for the
xyz positions directtly. I would assume that compass also works this way
and just spits back the corrected instrument readings for error-checking
purposes or just general interest.
In the casethat someone actually did write a nonlinear solver they would
almost certainly want to be able to store corrected instrument readings as
well as corrected xyz coords, for the reason that nonliniear solvers are
necessarily iterative and have a much faster convergence if they can use a
good guess for the starting point (and may not converge at all without a
good guess). Assuming the data haven't changed too drastically since the
last run of the software, then the previous solved values are likely to be
a good intial guess for this run, or at least be a better guess than the
raw measurements.
Lev
This archive was generated by hypermail 2b30 : Wed Jul 31 2002 - 23:00:00 CEST