Work

A Robust and Generalized Gauss-Seidel Solver for Physically-Correct Simultaneous Collisions

Year: 2025

Type: article

Abstract: Simulating multi-object collisions in real-time environments remains a significant challenge, particularly when handling simultaneous collisions in a physically accurate manner. Traditional Gauss-Seidel solvers, widely employed in physics engines, often fail to preserve the symmetry and consistency of multi-object interactions that are often observed in physics. In this paper, we present a generalized and robust Gauss-Seidel solver to overcome the difficulties in simultaneous collisions. Using spatial and temporal collision states to classify and resolve constraints, our algorithm ensures correct collision propagation and symmetry, addressing the limitations commonly found in existing solvers. Moreover, our algorithm can mitigate jitters caused by floating-point errors, ensuring smooth and stable collision responses. Our approach demonstrates fast convergence and improved accuracy in scenarios involving complex multi-object collisions.

Source: Proceedings of the ACM on Computer Graphics and Interactive Techniques

DOI: 10.1145/3728291

Authors: Huanbo Zhou, Zhenyu Xu, Xijun Liu, Xinyu Zhang

Institutions: East China Normal University, Shanghai Key Laboratory of Trustworthy Computing

Language: English

Topic: Computer Graphics and Visualization Techniques

Subfield: Computer Graphics and Computer-Aided Design

Field: Computer Science

Domain: Physical Sciences

Open Access status: bronze

Funder: National Science and Technology Major Project

Award: G6265082103