High-Fidelity Simulation of Human Muscles and Movement

点击翻译按钮获取中文摘要

Physics-based computer animation has improved our ability to simulate natural phenomena such as colliding objects, splashing waves, and moving characters. Although these simulation techniques can create compelling special effects for movies and games, the practical use of computer simulation in product design, ergonomics, and medicine remains very limited. In order for simulations to have significant impact beyond animation, they must not only be visually appealing, but also physically accurate and evaluated with experimental data. The proposed research program seeks to improve and evaluate the physical accuracy of human body simulations for applications in biomechanics. Our results will overcome the current limitations of human body simulation and improve our ability to diagnose movement disorders, guide rehabilitation, and train athletes. Two short-term research goals address fundamental limitations of current biomechanics simulations: simplistic visualizations and low-fidelity muscle models. We will create new techniques to visualize internal muscle deformations, realistic face animations, and exaggerated yet plausible character movements. We will also improve the fidelity of complex muscle models, such as those with feather-like structures, internal tendons, and broad attachment areas, to improve predictions of how muscle forces are generated and distributed. To verify our techniques and validate our results, our visualizations and simulations will be compared with measurements from real humans. This research will have considerable impact. The proposed projects will raise the bar for human simulation in terms of visualization richness, muscle detail, and evaluation methodology. Our computational methods and models will be widely disseminated through two world-leading, open-source software projects: ArtiSynth at UBC and OpenSim at Stanford University. In the long term, we aim to achieve a deeper understanding of human biomechanics that will greatly increase the predictive power of computer simulation for animation, training, diagnosis, and treatment planning.