No cookie for

A solid foundation to a bigger, better world

Whole-Earth simulation

I spent a few minutes pulling threads about whole Earth simulation out of Jim Kogler's mind. Here's what I learned:

There seems to be a general misunderstanding these days about the ability of game engines and military simulation systems to accurately represent the whole planet Earth in training simulations. Jim, why are people interested in modeling the whole Earth anyway?

As technology brings the world closer together, the range of our collective influence spreads wider. Military strategists are working through concepts for multi-domain operations that span the entire planet. Space is now a domain that contributes to our communications in the air, land, and sea domains. Increased use of autonomous sensor systems enable over-the-horizon intelligence, surveillance and reconnaissance. All of these applications depend on an accurate representation of our world — a big, round world — in multiple domains.

Why do I keep reading that Modeling & Simulation tools need to support a bigger world?

Because many ground-based training systems are built on commercial gaming engines that do their math in 32-bits. This presents a limitation to the mathematical representation of position in the simulation engine. Engines that compute coordinates as 32-bit values are somewhat limited to approximately 250 to 500km of space while maintaining a precision accuracy of about 3cm. Make your world much bigger than this and you've got accuracy problems that will prevent any meaningful simulation of the kinetic environment. While the details of this get a bit technical, it's important to note that using 64-bit location values allow for quite acceptable precision on the planet.

Many game engines have designed their environment to work on a flat earth reference system where "up" is aligned with the Z-axis. The lack of precision limits the size of the gaming area and the flat shape limits the accuracy of the environment model.

Why would game engines make these trade-offs?

For the most part, game engines were made to address smaller problems than the accurate representation of planet Earth. Game engines have always been focused on small areas, even if they seem really large. Game Engines have historically relied on hand made beautiful content, building top notch content is really expensive, making a big world just drives cost that doesn't make the game sell more. Plus, the math to move objects around in a flat earth is a lot easier than working in a geocentric one.

Well, games sure look like very real worlds.

Sure - and to get those amazing visuals, a tremendous amount of art needs to be developed and the position of the user needs to be controlled to stay inside the areas where that art is located. But great-looking content isn't the same as an accurate world. Games are optimized for effects. Physics is used to blow up objects or walls, simple bullet trajectories can be modeled, but there are no requirements for long range ballistics, anti-air missiles, or high release guided munitions that can have an operational range of 130km. The foundation of the game engine world is not solid when the scale increases. For games to get to an accurate planet-scale simulation, they would need to redo all their math in 64-bit resolution and move to a geocentric ellipsoidal world. That often involves throwing a LOT out and starting again.

 Then how is it that some games appear to simulate the whole Earth?

Sleight of hand! But game engines are not alone in this deception — many military modeling and simulation tools also employ a sleight of hand. You can claim a whole world simulation and then assume that users will plop down in one small spot on the planet and play on a flat plane tangent to the Earth near that spot. In this way, you can show a low-resolution model of the whole world but simulate in a small area as long as you don't move between several predefined tangential planes.

For example, you could create some entities in Boston or create some entities in Hong Kong — there isn't a problem. You just create a flat earth system tangential to the Earth and let those worlds play. If you are lucky, they aren't in the same system, and if you remain lucky, they don't need to interface with vehicles that are too far away from the tangential point. For ground-based training, it's seldom a worry. Vehicles in Boston aren't driving to Hong Kong, and it takes a long time to go 500km.

So, what is wrong with faking a big area with separate smaller ones?

Here's the rub. As training needs evolve, there is an increasing desire to move farther and farther from the origin of your little terrain patch. You don't just want to spin the globe and train there; you want to spin the globe and train there and communicate with something here — and those places can be really far apart. Look, today's battlefield is fully connected. Soldiers are communicating with pilots and pilots are communicating with satellite systems, and the whole electronic network could be vulnerable. If you really want to take modeling and simulation to the next step, you need to be able to model the whole thing, and all those pieces can be really, really far apart.

Without a strong foundation, you run into problems. The right way to solve this problem is to have a system that natively works in 64-bit math, where vehicles traveling between two distant points will move in the great-circle arc (which is the shortest distance between two points on a spheroid), maintaining full precision all the way around. 

How does MAK model the Earth environment?

We have been using 64-bit precision (i.e., double precision) on a WGS-84 ellipsoidal model for all the mathematical calculations within MAK ONE for about 20 years.Our customers have always been able to do things like having a plane take off and fly continuously around the world without shifting or offsetting coordinate systems while simultaneously having human characters walking around accurately on the ground on the other side of the Earth. In recent years we have added satellite movement - you can load a whole satellite constellation by importing TLEs (Two Line Elements) which accurately describe their movement. A big earth simulation shouldn't end at the end of the atmosphere, electronic and cyber threats don't end where our eyesight runs out.

Why did MAK choose the bigger, more accurate path?

We did this because we built MAK ONE from the ground up to be a Multi-Domain Environment. MAK ONE customers simulate everything from the bottom of the ocean to way out in space and everywhere in between. The real world is large, and our nations' threats are big, distributed, and scary. Modern simulation engines must care about the whole threat environment, as well as the whole joint forces environment.

For example, the difference in the effectiveness of a radar system placed on the coast at sea level is radically different from the effective range if placed on mountains 2,000 feet up. Building an Integrated Air Defense simulation system that considers terrain like mountains is good, but until it considers the curvature of the earth, it's not accurate.

Okay Jim, any parting thoughts?

The MAK ONE environment provides that solid foundation for you to build your world, whether you need a small area, over the horizon, or the whole planet. You keep building and know that with MAK ONE, you are building on a solid and proven platform for over 30 years. It's a big world out there already. We have your back — go play in it.

Tech Tip: Manage the flow of training exercises wi...
MAKer Spotlight: Alicia Combs, Director of Program...

Related Posts

By accepting you will be accessing a service provided by a third-party external to