Re: Funding by DOD for TerraGear related challenge

Anyone want to explain this to me?  I'm not sure I agree with (or
understand their assertions and/or terminology.)  Is there any chance this
isn't already targeted towards someone else?  Sounds like someone wants to
get paid to develop or enhace a CLOD/ROAM algorithm and got the military to
write up a project request for them.

Alex Perry wrote:
> ------------------------------------------------------------------------
> 
> 
> Subject: [Terragear-devel] Funding by DOD for TerraGear related
> challenge From: Alex Perry <[email protected]> Date: Sun, 11 Jan 2004
> 12:58:24 -0800 (PST) To: [email protected]
> 
> 
> [apologies for formatting] 
> http://www.acq.osd.mil/sadbu/sbir/solicitations/sbir041/af041.htm
> 
> AF04-064 TITLE: Algorithms for Run-Time Terrain Deformation
> 
> TECHNOLOGY AREAS: Information Systems
> 
> OBJECTIVE:  Develop algorithms and prototype minimum hardware and system
> requirements for implementation of run-time deformation of terrain based
> on elevation of topographic features in real-time visual simulations.
> 
> DESCRIPTION:  Current image generator architectures force topographic
> features to conform to an underlying terrain model.

I wonder what image generators they have in mind.  I wonder what underlying
terrain model they are thinking of?  I wonder what topographic features
they want to impliment that current terrain models can't handle?

> This IG technique
> is backwards in terms of training value, because the three-dimensional
> position accuracy of topographic features is of greater interest to
> pilots versus the general accuracy of a featureless terrain skin.

So are they saying they want to be able specify certain areas with ultra
accuracy, rather than having the entire earth covered with so-so accurate data?

> A
> superior IG architecture should necessarily deform terrain in accordance
> with the elevation values of topographic features.

Can anyone explain what they mean by terrain deforming?  If the terrain is
already accurate enough, why does it need to be deformed.  It seems like
they must have something in mind here that they aren't fully sharing with us.

> The ideal architecture would perform deformation at run-time.

I'm guess this is ideal for the group that is already lined up to get this
contract. :-)

> The run-time
> deformation of terrain controlled by the z elevation of topographic
> features should dramatically reduce the cost and time to produce terrain
> databases for training and simulation.

Now this assertion is completely beyond me.  They've completely lost me.

> The cost of generating terrain
> databases is steadily becoming a larger percentage of the total cost of
> procuring training devices.

I do very much believe this assertion.

> This technique could help alleviate the
> cost growth problem of terrain databases, as well as help reduce the
> time it takes to generate databases for mission rehearsal purposes.

So they believe that using some sort of ROAM/CLOD algorithm that dynamically
computes the terrain on the fly will save money in database generation costs?
Are they serious?  What am I missing here?

> PHASE I:   Identify existing and develop new algorithms/techniques to
> perform the real-time terrain deformation function.  It will establish
> the minimum hardware and system requirements for implementation of each
> algorithm/technique.  Finally, the associated performance strengths,
> weaknesses, risks and performance/cost tradeoffs associated with each
> algorithm/technique will be evaluated.

Sounds like Ben Disco could submit his VTP project here and get the cash
straight up without doing any additional work.

> PHASE II:   Select and prototype one or more of the most promising
> candidate algorithms/techniques to work with commercial IG hardware.

I wish this was written so that the bid winner would have to demonstrate
how their new approach to modeling terrain so significantly reduces
database creation costs, rather than just buying all these strange
assertions up front.

Sounds like a cream puff project in the sense that there's no consequence
for being stupid.  Evaulate current algorithms, wave your hands and pretend
to come up with a few new wrinkles on the existing algorithms, pick one,
prototype it ... sounds like busywork to me. :-)  Almost sounds like
someone who doesn't fully understand the issues of terrain modeling is
getting hornswaggled by someone who wants to do a project not requiring a
lot of thought.

Or I could be miss understanding something here ... what do I know?
They'll flounder around with this stuff for a while and eventually come
crawling back to terragear. :-)

> DUAL USE COMMERCIALIZATION:  The following are a few of many potential 
> Dual Use applications of this technology: 1. Mission planning
> visualization. 2. Embedded, counter-measure resistant, stealth
> navigation for UAV's, RPV's and piloted vehicles. 3. Virtual
> landscaping. 4. Visualization for large scale, cultural construction
> projects.
> 
> REFERENCES:  1. Hakl, Henri.  "Diamond Terrain Algorithm: CLOD for
> Height Fields," University of Stellenbosch, 2001. 2. Röttger, Stephan,
> et al.  "Real-Time Generation of Continuous Levels of Detail for Height
> Fields," University of Erlangen-Nuremberg, 1998. 3. Duchaineau, Mark, et
> al.  "ROAMing Terrain: Real-time Optimally Adapting Meshes," IEEE
> Visualization '97 Proceedings, 1997. 4. Lindstrom, Peter, and Pascucci,
> Valerio (Lawrence Livermore National Laboratory).  "Visualization of
> Large Terrains Made Easy," Proceedings of IEEE Visualization 2001,
> October 2001, pp. 363-370, 574. 5. Zhao, Youbing, et al.  "A Fast
> Algorithm For Large Scale Terrain Walkthrough," International Conference
> on CAD & Graphics 2001, China.
> 
> KEYWORDS: Training Simulation, Visual Simulation, Image Generation

-- 
Curtis Olson   Intelligent Vehicles Lab         FlightGear Project
Twin Cities    [email protected]                  [email protected]
Minnesota      http://www.menet.umn.edu/~curt   http://www.flightgear.org