RadiusFPS: Efficient Farthest Point Sampling on CPUs and GPUs via Spherical Voxel Pruning

2026-06-04

AI Summary

This work addresses the high computational complexity of traditional farthest point sampling (FPS), which hinders its deployment in robotic systems requiring real-time performance under limited computational resources. The authors propose RadiusFPS, a novel framework that introduces spherical voxel pruning to preserve standard FPS sampling quality while eliminating redundant distance computations via geometric conservative bounds and reducing update overhead through coordinate-level skip tests. Furthermore, they design RadiusFPS-G, a GPU warp-level fused implementation that integrates voxel selection, pruning, and distance updates to significantly minimize memory accesses. Experiments on S3DIS, ScanNet, and SemanticKITTI demonstrate that RadiusFPS-G achieves up to 2.5× speedup over existing GPU-based FPS methods, uses approximately half the memory of QuickFPS, maintains comparable segmentation accuracy, and delivers the fastest end-to-end inference latency to date.

Abstract

Point clouds are a primary sensory representation for robotic perception, underpinning LiDAR-based autonomous driving, simultaneous localization and mapping (SLAM), and navigation. Within these pipelines, Farthest Point Sampling (FPS) is the most well-known downsampling operator, as its uniform coverage preserves the geometric structure on which downstream perception relies. However, the large time complexity of classical FPS scales poorly with the million-point-per-second rates of modern 3D sensors, making it a dominant latency bottleneck that conflicts with the real-time and limited onboard compute budgets of robotic systems. Therefore, we propose RadiusFPS, an FPS acceleration framework based on spherical voxel pruning that preserves the standard FPS update rule under the same initialization and tie-breaking policy. By indexing the point cloud with spherical voxels, RadiusFPS derives a conservative geometric bound that prunes redundant distance computations in each iteration, complemented by a coordinate-wise point-skip test that removes residual updates. We further introduce RadiusFPS-G, a warp-level GPU implementation that fuses voxel selection, pruning, and distance update into memory-coalesced kernels, eliminating costly global-memory round-trips. On indoor (S3DIS, ScanNet) and outdoor LiDAR (SemanticKITTI) benchmarks, RadiusFPS-G attains up to 2.5x speedup over GPU-based FPS and matches or exceeds QuickFPS among the evaluated methods while using roughly half its GPU memory, with comparable segmentation accuracy. When coupled with the learning-based FastPoint sampler, the resulting pipeline achieves the fastest End-to-End inference among all evaluated configurations. These properties make high-quality FPS-style sampling practical for latency- and memory-constrained robotic vision.

Problem

• Farthest Point Sampling
• point cloud downsampling
• real-time robotics
• computational latency
• 3D perception

Innovation

• Farthest Point Sampling
• Spherical Voxel Pruning
• GPU Acceleration
• Point Cloud Downsampling
• Memory-Coalesced Kernels