Broad Phase Collision Detection: New Methodology and Solution for Standardized Analysis of Algorithms

Abstract

Collision detection is a computational problem focused on the identification of geometric intersections between objects and, in general, proximity relationships among them. Despite its notorious relevance and applications in various computing fields, few authors have proposed solutions that are both general and scalable. Additionally, until the time of publication of the results of this work, there was no standard methodology for the analysis of algorithms, neither in academia nor in the industry: only proprietary scenes and comparative studies had been developed, making it difficult to reproduce and compare results. To tackle the issues previously mentioned, we present a new general and scalable solution for broad phase collision detection and a new methodology for comparative analysis of algorithms, named Broadmark, whose open-source code is publicly available, with the goal of transferring knowledge to academia, industry, and society, so far lacking in the scientific literature. Thus, by doing so, we aim to contribute to the generation of robust and multi-faceted solutions applied to various scenarios and, consequently, to greater transparency, ease of modification/extension and reproducibility of results.1