Take one of those ubiquitous white plastic lawn chairs. Pound it. Twist it. Toss it against the wall. Sure, it will bend, but no matter what you do short of breaking it, the thing still looks like a chair.

	Bill Wade, Post-Gazette Doug James in his office at Carnegie Mellon University. James was named one of Popular Science's "Brilliant 10" young scientists. Click photo for larger image.

It's a simple enough observation but, for Dr. Doug L. James, it turned out to be particularly keen.

An assistant professor of computer science and robotics at Carnegie Mellon University, he used it to develop ways to make computers simulate collisions that is a thousand times faster than previous methods.

And, of course, he used his method to simulate collisions between white plastic lawn chairs -- about 3,600 of them, all falling across a computer screen in a cascade that resembles a waterfall.

That he could do this simulation in a few hours, rather than a couple of months, has drawn attention to the 33-year-old scientist from such groups as Pixar, the computer animation house. This month, Popular Science magazine named him one of its "Brilliant 10," a list of largely unknown but innovative young researchers.

Though these methods have obvious applications to computer-generated animation and video games, they also could address a wide variety of problems.

"You can use it to help a robot build a model of his shoelaces and [enable him to] tie his shoes," said Dr. James, a mathematician and computer scientist. They might also be used to develop realistic scenes for surgical simulations and might even be used for drug design.

In a computer, the shape of an object is typically represented by tens of thousands of tiny triangles. In conventional programs, when an object collides with something, the shape of each triangle is recomputed, based on physical principles.

Bill Wade, Post-Gazette The plastic lawn chair simulation Click photo for larger image.

That requires a massive amount of computing, which means many such graphics programs run very, very slowly. To speed things up, Dr. James said, "you need to be able to cheat in some way."

That's where the lesson of the lawn chair comes in. Like a plastic lawn chair, most things aren't made of clay. They aren't endlessly deformable, so there is a limited number of shapes that can occur, what might be called a "shape space."

That means a computer only needs to pay attention to a handful of points to understand the object's shape and orientation and the computer program needs pay attention only to those triangles that are actually touching.

"In most cases, things don't touch each other all over their surfaces," he explained. So, using what he calls "bounded deformation trees," the computer does detailed computations only for those triangles that are touching. That reduces the amount of computing drastically and thus speeds up the entire process.

Dr. James, who joined Carnegie Mellon in 2002 after earning a doctorate degree in computer science at the University of British Columbia, has applied the technique to animals, simulating them as they run and collide.

Though the computer graphics community is understandably interested in the techniques, his own interests have more to do with using them to model the real world, both for helping robots navigate and manipulate appendages and for simulating surgical operations.

"My main interest now is understanding how far we can push this," he said, and particularly looking at how to complement the collision simulations with equivalent technologies for sound and touch. "Can you make a virtual environment that is truly immersive?"