Real-time collision culling of a million bodies on graphics processing units

Abstract

Figure 1: GPU-based collision culling for massive bodies. N -body collision detection for 1M arbitrarily moving boxes (first image), realtime simulation of 0.3M particles of random size on GPUs (second and third images), real-time rigid-body dynamics for 16K torus models of varying sizes on GPUs (fourth and fifth images). In these challenging benchmarks, our algorithm can find all the colliding bodies at interactive rates. AbstractWe cull collisions between very large numbers of moving bodies using graphics processing units (GPUs). To perform massively parallel sweep-and-prune (SaP), we mitigate the great density of intervals along the axis of sweep by using principal component analysis to choose the best sweep direction, together with spatial subdivisions to further reduce the number of false positive overlaps. Our algorithm implemented entirely on GPUs using the CUDA framework can handle a million moving objects at interactive rates. As application of our algorithm, we demonstrate the real-time simulation of very large numbers of particles and rigid-body dynamics.

A hybrid fixed-function and microprocessor solution for high-throughput broad-phase collision detection

“…For the GPU comparison, we selected Bullet's GPU-based sweep and prune, for which we recorded the time spent checking for collisions between objects, while excluding the transfer of data between the CPU and GPU. It is an implementation of the design proposed by Liu et al [14]. The execution times of all these systems are tabulated in Table 4, and the acceleration factors of our system relative to the CPU and GPU are plotted in Fig.…”

“…As GPUs developed fully programmable cores, researchers moved to utilise these. Liu et al [14] outline a broad phase that represents objects as a collection of spheres processed using spatial partitioning, followed by full-sort sweep and prune along a single axis chosen to minimise the number of overlaps. The narrow phase is avoided using a penalty algorithm for rigid-body dynamics, although this can introduce substantial divergences from expected results.…”

“…One solution would be to switch from incremental to full-sort sweep and prune, but this essentially obviates the algorithm's primary advantage of coherence. The GPU sweep-andprune implementations designed by Liu et al [14] and Avril, Gouranton and Arnaldi [15] represent an attempt to trade coherence for parallelism. Despite the promise shown, full-sort sweep and prune would not be entirely amenable to microarchitectures as it makes significant use of sorting.…”

Parallel GPU-based collision detection of irregular vessel wall for massive particles

“…However, these algorithms mainly focus on the detection between a small number of objects [8], and it is not suitable for fast collision detection that contains massive objects, especially massive particles [9].…”

“…The reason to use uniform grid is because it is simple and similar for each grid, hence, it is suitable for GPU implementation [9]. As shown in Fig.…”