In his distinguished career in computer science, Demetri Terzopoulos has evolved from simulating
real-world objects, to creating comprehensive simulations of reality.
In the late 1980s, while working as a scientist at a Schlumberger Research center in Silicon Valley, Terzopoulos helped pioneer physics-based computer graphics techniques to simulate how cloth and other nonrigid objects move, such as a flag flapping in the wind.
These techniques were soon incorporated as special effects in Hollywood films. So today, whether it’s a completely computer-animated movie like Toy Story or Monsters, Inc., or a film that incorporates computer generated effects like the Harry Potter or Star Wars franchises, Terzopoulos’ pioneering work helped set the stage for some of those films’ amazing images and special effects. And this work earned Terzopoulos a 2005 Academy Award for Technical Achievement.
Terzopoulos, Chancellor’s Professor of Computer Science, has been working on the profound challenge of realistic human simulation.
This includes emulating complex human activity in urban environments, which has resulted in computer models featuring autonomous virtual pedestrians. These models are more than simple crowd simulations; rather, they are comprehensive artificial life models of individual human appearance, locomotion, behavior, and cognition. Each virtual human can independently sense and interpret the environment, then make decisions based on several different factors, and perform the appropriate actions.
This work has applications not just in the motion picture and interactive game industries, but also in simulating the behavior of groups of people in different settings—for example, predicting what a crowd in a train station will do in some specific emergency situation.
“We have already simulated some nontrivial social interactions among pedestrians through the use of decision networks,” said Terzopoulos, adding that “There remain many fascinating open problems to investigate, but as a next step, we hope to model some relevant aspects of human verbal communication so that the social interactions of our virtual pedestrians will be much more lifelike.”
Terzopoulos, his Ph.D. student Sung-Hee Lee, and postdoctoral researcher Eftychios Sifakis, have also developed and continue to refine a comprehensive biomechanical model of the human body. This computer model includes a skeleton with 75 bones and 165 jointed degrees of freedom, plus a staggering 846 muscle actuators. Surrounding the muscleskeletal framework is a 3D finite-element mesh that simulates the realistic deformation of soft tissues.
The highly detailed model, for which Lee received the 2009 UCLA Computer Science Outstanding Ph.D. Award, can potentially facilitate advances in medical technology or surgical techniques, among many other applications.
According to Terzopoulos, enabling such an elaborate biomechanical model to control itself remains a big challenge. “We have successfully developed a neuro-muscular controller that can be trained to actuate the neck muscles so as to balance the head in gravity atop the cervical column and to synthesize voluntary head movements that are essential to many aspects of human behavior,” he said. “And we are now endeavoring to generalize our approach with the hope of controlling the order-of-magnitude greater number of muscles in the complete biomechanical body model.”
So, what is the big picture in realistic human simulation? “My long-term objective is a computer simulated world that approaches the complexity and realism of the real world, inhabited by virtual humans that look, move, and behave as much as possible like real people,” said Terzopoulos. “Such a reality emulator could be used in revolutionary ways across multiple scientific disciplines.”
Terzopoulos’ ongoing work on realistic human simulation was recognized in April with a 2009 Guggenheim Fellowship.