AI Approach Allows Computing Game Physics In Microseconds
Check out a novel approach that uses data-driven methods for physical simulation.
The approach was discussed in the paper "Subspace Neural Physics: Fast Data-Driven Interactive Simulation" by developers from Ubisoft. First, start by watching a quick overview shared by the Two Minute Papers channel above.
The team states that such a method lets developers trade precomputation and memory footprint in exchange for improved runtime performance. Their method mixes subspace simulation techniques with machine learning which to get an efficient subspace-only physics simulation that supports interactions with external objects.
"We also present an interpretation of our method as a special case of subspace Verlet integration, where we apply machine learning to efficiently approximate the physical forces of the system directly in the subspace," states the paper's abstract. "We propose several practical Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page."
Basically, the new approach is all about saving time on repetitive tasks when working on games. You can find the full paper here. Don't forget to discuss the method in the comments.
