CAE's Common Database, a plug-and-play database that can move across simulators, was originally developed for U.S. special operations forces. (CAE)
Simulation can be sexy: the excitement of explosions as a Humvee navigates a market, smooth barrel rolls in a jet flying over sprawling Italian countryside, immersive worlds for deskbound target practice. But most people’s eyes stop at the visuals and forget what lies beneath. All of the beautiful rendering in the world doesn’t do a lick of good without the data to power it.
“We’ve been spending a lot of time on our databases,” said LeAnn Ridgeway, vice president of Rockwell Collins’ simulation and training group. “You can have a great image generator that would show you the most beautiful realism in the world, but if you haven’t kept your database realism up to speed, you’re going to lose that.”
And as the number of simulations and virtual training grows, governments and companies face a common challenge: determining how to untangle and organize the petabytes of information that feed their machines, share it (when they want to) and keep from wasting resources by duplicating information that already exists.
Creating Virtual Worlds
Different databases power simulations in different ways. Some of the best-known are the terrain databases that feed simulators elevation and geographic data, the physics databases that allow for real-world interactions, and databases of heat profiles that allow for training with night-vision goggles.
However, there are also database systems that organize models that can be used to create virtual worlds by independent modelers or in programs such as Second Life. While various commercial companies have started aggregating these models, the U.S. military is also growing its models database with the 3D Repository.
Created by the Department of Defense’s Advanced Distributed Learning initiative and first demonstrated at the GameTech 2010 conference, the 3D Repository is an open-source collection of models that developers can use to populate their scenarios.
ADL supervisor Frank DiGiovanni said the main impetus behind the project was to give the DoD training community a way to find and share models more easily, without having to do an exhaustive search.
“This is a really great resource to cut down on duplicative efforts and also let our community see what’s available,” said DiGiovanni, who also serves as the director of training readiness and strategy in the Office of the Deputy Assistant Secretary of Defense for Readiness.
The variety in the 3D Repository is unexpected. There are plenty of military-specific objects, such as a Bell Augusta rescue helicopter, the Oshkosh M-ATV, an MP5 submachine gun and a Buffalo mine-protected vehicle. But there are also plenty of everyday objects: magazine stands, kitchen tables, warehouses, newspapers, even pool cues and easy chairs.
“When you build a training environment, you need a realistic one,” DiGiovanni said. “So it’s not always going to be military hardware. And we have a lot of people who build training scenarios using a virtual approach to training.”
At this point, many of the models are just artistic representations, rather than fully physics-based or kinetic models. But DiGiovanni said the combination of the two is useful for training.
“A lot of these things provide the fidelity necessary, but the 3-D physics-based models are still going to be required,” he said. “And eventually we could see some of those being loaded up on the site as well.”
ADL was serving roughly 450 users as of September, according to ADL director Kristy Murray. Among them are the Army Research Lab, the Naval Postgraduate School and several small companies.
“All four of the services use it,” Murray said. “That doesn’t mean it’s their main, go-to place.”
Among the 3D Repository’s key abilities is converting models to various file types. A file uploaded to the repository is translated to Collada, the default format. When a user downloads the model later, it can be “written” into any requested format. There are options to convert between many common files, including Autodesk 3ds and FBX, Wavefront OBJ and Google Sketchup.
And there’s one more trick up the repository’s sleeve: It can link up with other 3-D repositories.
“This has an ability to federate with other repositories, and that’s the real power,” said Murray. “Not only taking the 3-D models that are in there, but then being able to talk to other repositories where you might be able to take other resources.”
With Sims, Sharing is Caring
The ADL is an example of the kind of database sharing and interoperability that the military is pushing toward. The Defense Department accepted recommendations from a July Government Accountability Report that suggested the Air Force “develop databases to provide a common constructive environment” and help different simulators translate data.
“One of the things we focus on in this office is joint training,” DiGiovanni said of his department. “So if the Army builds an M1A1 Abrams tank, it would be useful to allow the Air Force to use that.”
While there are still thousands, if not hundreds of thousands, of models missing from the 3D Repository, the intention is right. Without a common, shared compendium of models, data can end up trapped behind firewalls in different branches of the military.
“They have to go talk to the owner of the model, and then negotiate getting the kinetic model and then put it into their system,” DiGiovanni said. “Now, they can go to a central website, and these things are available for them to very easily download.”
ADL’s charter encourages the federal government to use shared data in training and education. The open source formatting, in particular, can be a foreign concept to a military steeped in protections and control.
“What’s interesting is, the industry has gone toward this open source, crowdsourcing business model,” DiGiovanni said. “It’s kind of ground-breaking for the Department of Defense.”
But open sharing of databases may be the key to providing the interoperability that makes simulations play nicely together. Shared databases mean matching simulations with correlated data. So when individuals conduct simulations together, the worlds they experience will match up exactly, adding to the realism and fluidity of the training. Anything from using different databases or even outdated sets of information can alter the training experience.
“The common database becomes a fundamental building block,” said Gene Colabatistto, head of flight simulator CAE’s military training division. “And at the end of the day, it’s what everybody hooks into. And if you solve that problem, you make the whole architecture of the system a lot less expensive.”
One potential solution to differing databases is sharing the data sets through the cloud. If the information is available for anyone training to dip into, then parties training together are guaranteed to be working from the same map.
When it comes to military training, data that matches the real world is vital, Ridgeway said. Everyone wants real-time data, rather than spending weeks building a database that looks real only to have it changed by a natural disaster or other impact.
“Real-time imagery ported immediately into your simulators — I think that’s where it’s going eventually,” Ridgeway said. “The technology is rapidly developing. I think in five years, that’s where we’ll be. We’ll be seeing real-life scenes in LVC simulations.”
A final challenge facing the databases that power the sims of the future? Making sure various “layers” of databases can talk to one another. For example, a bomb might make a crater in a simulator, but a tank driving over that crater isn’t likely to dip down when it rolls over it.
“There’s no interaction between the layers,” Colabatistto said. “It’s a system engineering problem that just hasn’t been solved yet.”
As the weapons and simulators become more complicated and advanced, so too must the databases that power them.
With live, virtual and constructive training becoming more common, making sure the databases that power the three types of training match could be the difference between successful home station training and failure.