PEARS - Physics Extension And Representation through Semantics
IEEE Transactions on Computational Intelligence and AI in Games, 8 (2):
178-189 (июня 2016)DOI: 10.1109/TCIAIG.2015.2505404
Аннотация
Today\u0027s physics engines mainly simulate classical mechanics and rigid body dynamics, with some late advances also capable of simulating massive particle systems and some approximations of fluid dynamics. An accurate numerical simulation of complex non-mechanical processes in real-time is beyond the state-of-the-art in the respective fields. This article illustrates an alternative approach to a purely numerical solution. It uses a semantic representation of physical properties and processes as well as a reasoning engine to model cause and effect between objects, based on their material properties. Classical collision detection is combined with semantic rules to model various physical processes, e.g., in the areas of thermodynamics, electrodynamics, and fluid dynamics as well as chemical processes. Each process is broken down into fine-grained sub-processes capable of approximating continuous transitions with discretized state changes. Our system applies these high-level state descriptions to low-level value changes, which are directly mapped to a graphical representation of the scene. We demonstrate our framework\u0027s ability to support multiple complex, causally connected physical and chemical processes by simulating a Goldberg machine. Our performance benchmarks validate its scalability and potential application for entertainment or edutainment purposes.
