Newsgroups: comp.robotics
Path: brunix!sgiblab!swrinde!cs.utexas.edu!howland.reston.ans.net!europa.eng.gtefsd.com!MathWorks.Com!news.kei.com!ub!netfs.dnd.ca!dres.dnd.ca!dres.dnd.ca!sbogner
From: sbogner@dres.dnd.ca (Stephen Bogner)
Subject: Panoramic Vision Systems for Teleoperated Vehicles
Message-ID: <sbogner.34.2DB6FF1B@dres.dnd.ca>
Summary: Reposted from sci.military
Lines: 111
Sender: news@dres.dnd.ca
Nntp-Posting-Host: steveb
Organization: Defence Research Establishment Suffield
X-Newsreader: Trumpet for Windows [Version 1.0 Rev A]
Date: Thu, 21 Apr 1994 22:33:00 GMT
Lines: 111

Here is a repost of an article that I submitted on a related thread a few days 
ago.  It may be relivant to you....

From: sbogner@dres.dnd.ca (Stephen Bogner)
Subject: Panoramic Vision Systems for Teleoperated and Armoured Vehicles
Summary: Vehicle Concepts Group is pushing for a project in this area
Date: Tue, 19 Apr 1994 16:14:18 GMT



>From sbogner@dres.dnd.ca (Stephen Bogner)

To follow up on some of the recent discussions about vision systems in Re: 
Merkava and Re: Tank Commander Vision Systems...

I have been pushing, for about 6 months or so, a project to implement such a 
panoramic vision system onto a teleoperated vehicle, with the expectation that 
it would be fitted onto some of our conventional vehicles, especially armoured 
vehicles, if it proves out.  There are some very interesting ideas that have 
been raised in recent discussion on this news group, and so I think that it 
would be beneficial to share with sci.military some general information about 
the technical approach that I am taking, and invite your comments.

The primary objective is to increase the level of situational awareness of 
the teleoperated vehicle operator, or the armoured vehicle crew, by providing 
a 360 degree panoramic view and sonic monitoring of the environment around the 
vehicle.  Secondary benefits are expected to include the elimination of the 
conventional pan-tilt mechanism from robotic vehicles (an important 
simplification of such systems, since these mechanisms tend to be very 
problematic in practice), and the possibility of improved communication 
amongst the AFV crew if they can reference a common exterior view, and 
improved safety if the driver can actually see  more than the 20-30 degrees 
available to him now.  Many other advantages will be obvious to this audience.

Previous attempts to create such systems have focused on multiple cameras and 
multiple monitors, or some form of rotating camera.  While they have all 
achieved some success, these techniques have had various drawbacks that 
have prevented their serious use.  Several panoramic lenses have been built 
over the past few years, the most promising of which is the Panoramic Annular 
Lens (PAL) invented by a Hungarian (Greguss) several years ago.  In my view, 
this is likely the most promising lense for this type of application, and I 
would expect to see capable panoramic viewing systems built around this lense 
begin to emerge in the next few years.  However, the approach that I am 
considering promises to be extreamely simple to implement, and should be even 
more economical than the PAL.

A conical mirror (fabricated perhaps from stainless steel, using a simple 
lathe and polishing pastes) is centrally mounted above the barrel and lens of 
a vertically directed conventional camera, and the reflected image of the 
surroundings is captured on the video CCD (or film, if you prefer) of the 
camera.   The conical mirror acts as a "remapper" and converts the spherical 
world image coordinates (r, alpha, beta; where r = depth of field, say 4m to 
infinity (remembering that an image in a plane mirror is virtual, and so the 
distance to the image for focal length calculations is distance lens-mirror 
plus distance mirror-object); alpha = 0 to 360 degrees;  and beta = some range 
of angle depending upon the angle of the mirror cone,  to polar coordinates 
(R, theta) on the CCD or film, where R maps to beta and theta maps to alpha.   
This image looks like a clock view, with the horizon forming a ring around the 
centre, the sky being on the outside edge.  The front of the vehicle is at the 
12 o'clock position, where the image is erect, the rear of the vehicle is at 6 
o'clock, where the image is inverted.  There is some uniform curvature in 
theta (due to the mapping), but there is no distortion in R.  The remapped 
image, whether directed through IR, NOD, or passed through light piping, is 
captured in digital form, and processed to allow such things as false color 
enhancement of "features of interest", HUD type overlays, and so on.  A 
digital image can also, of course, be manipulated for pixel zoom of an object 
of interest, automated tracking of identified targets, assisting in ballistic 
calculations, and so on.  The mirror is attached by "legs" to a male-threaded 
ring that threads directly onto the camera barrel using the female-threads 
provided for installing light filters.  

The remapped, manipulated, enhanced image may be presented to the 
operator/crew in several ways:

1.  The polar image may be presented directly to the operator.  

2.  The polar image may be projected onto a conical mirror that 
"undoes" the initial remapping, and reflects the image onto a 
spherical/toroidal screen, giving a type of Virtual Reality, helmet-less heads 
up display.  (One must not forget to address the issues of luminous flux in 
projection...)

3.  The polar image can be remapped into cartesian coordinates for display on 
a conventional (or custom) monitor.  (While I have not done this myself, I 
understand that there are several video chips and some software "morphing" 
packages that can do this type of thing relatively painlessly.)  With a 360 
degree alpha and 60 degree beta range, the 6:1 aspect ratio could be remapped 
rather nicely onto two 180 degree wide strips on a conventional monitor, 
leaving a bit of space for tactical information and perhaps a zooming or 
gunnery fire control window.

The system has no moving parts, is composed of simple and inexpensive 
components which are either readily available from commercial sources or may 
be fabricated by any competent metal-working shop, and has obvious 
utility in applications beyond vision systems for teleoperated robotics or 
armoured vehicles; for example in the home entertainment field.

I am particularily interested in comments regarding the operational utility 
and the technical aspects of this approach.  As I am still agitating for 
support (and funding) within my own organization, I would also value 
expressions of interest from agencies, commercial or governmental, who are in 
a position to collaborate in this project.

Regards, Steve.


.............................................................................
Stephen Bogner  (DRES/DTD/MES/Vehicle Concepts Group)     sbogner@dres.dnd.ca
(403) 544-4786  DRE Suffield; Box 4000; Medicine Hat, Alberta; Canada T1A 8K6
"Always leave your clothing and weapons where you can find them in the dark."
                                         - from the notebooks of Lazarus Long
