CS Colloquium: Physical Motion Control and Analysis in Games, Visual Effects and Training.

Monday, March 26, 2012
ENGR 4201
Brian Allen


The synthesis of realistic motion is a key component of visual effects and computer games. Correspondingly, the dual problem of recognizing a particular grace in motion, has the potential to improve the training of movement skills and animation. As humans become proficient with a manual skill, their motion becomes more fluid, more efficient and more compliant. Physical simulation, now cheap and ubiquitous, is a promising means for creating and understanding motion. In contrast to key-frame animation or motion capture, characters driven by physical laws can move in new, dynamic and unforeseen ways in response to their environment and user interaction. However, a key challenge with using physically simulated characters is developing controllers capable of reproducing the fluidity and compliance of well-practiced motion.

In this talk, I will present new approaches for both the control and the analysis of fluid and compliant physical motion. For control, I will introduce a novel solution to a classic, low-level control equation. This solution provides an analytic method for determining the character's compliance, that is, how the simulated character will respond to unexpected collisions. I will also introduce a biologically inspired method for generating high-level controllers capable of complex and dynamic whole-body behaviors.

Additionally, I will show that such control techniques can also serve in the analysis of human motion, for example, in estimating motor-skill level based only on observed motion, or in predicting future movements. I will illustrate both synthesis and analysis of motion with examples from a range of applications in computer games, visual effects, robotics, virtual reality and medical training.

Speaker's Bio

Brian F. Allen is a Senior Research Fellow at the Institute for Media Innovation at the Nanyang Technological University in Singapore. His research focuses on natural motion and physical simulation with applications to games, visual effects and medicine. He has ten years of software development experience, including working with Industrial Light and Magic R&D, University of Southern California's Institute for Creative Technologies, and co-founding and serving as CTO of Silicon Age, a San Francisco-based software consultancy. He received his B.S. in Computer Science from Iowa State University and his PhD from the University of California, Los Angeles with the advisement of Petros Faloutsos.