Motion Simulation of Geometric Constraint Structures
Files
Date
2012-05-11
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Motion simulation is a classical problem in areas of research such as CAD (Computer Aided Design), robotics, and protein folding and flexibility. Efficient motion simulation techniques could facilitate advancements in many such domains. Improvements in CAD motion simulation would offer mechanical engineers more quantitative feedback about the components of their mechanisms resulting in faster development. A model of protein motion and flexibility could help predict how molecules will interact, which may be able to considerably reduce trial and error for researchers attempting to design drugs that could ameliorate or cure various diseases.
While sophisticated motion simulation techniques exist, most have computational limitations. FEA (Finite Element analysis) allows engineers to study structural performance of their mechanisms. However, the growing complexity in models requires newer techniques for motion simulation. Molecular dynamics is widely recognized as the most accurate method for simulating the motion of proteins; however, similarly, it is computationally very expensive and, in some cases, requires prohibitive computing power and resources. Rather than relying on current, computationally expensive tech- niques, we worked to help distill the intricacies of structural motion by using geometric constraints to model structural interactions. This level of abstraction has facilitated development of more efficient motion simulation techniques, through incorporating methodologies from other domains, such as the video game industry. In particular, we focused on generalizing the approach of Ragdoll physics.In this thesis, we will describe the software tool we developed to simulate the motion of geometric constraint structures and new strategies for doing so. We will also explain and analyze the experimental results from applying our different techniques and comparing them to some cur- rent methods, specifically those of Ragdoll physics. For certain strategies and classes of structures, we found that our techniques outperform existing ones. These results indicate that our work may lead to algorithms that improve simulation for molecular motion.
Description
Keywords
Motion Simulation, Bar-and-joint structures, geometric constraints