{"id":"0699654a-d315-4bf9-a858-0e18cbfb5f98","arxiv_id":"2606.06255","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"RadiusFPS accelerates standard FPS via spherical voxel pruning and a GPU kernel, delivering up to 2.5x speedup on indoor/outdoor benchmarks with comparable accuracy and lower memory use than prior methods.","lead":"RadiusFPS speeds up farthest point sampling for large point clouds by using spherical voxels to skip unnecessary distance calculations while keeping the original FPS selection rule. Robotic systems processing LiDAR data could run perception pipelines faster on limited onboard hardware.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Pruning bound conservativeness unverified against exact FPS equivalence","rationale":"The reader's weakest assumption directly identifies the unverified safety of the geometric bound; confirming exact output equivalence on real data would resolve the UNVERDICTED status without altering the current verdict.","tokens_in":1795,"tokens_out":241,"duration_ms":20334,"concrete_test":"Run both naive FPS and RadiusFPS on the same 10k-point subset of S3DIS (fixed seed, identical initialization and tie-breaking) for 1024 samples; confirm that the exact sequence of selected indices matches on every run.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The speedup, memory, and accuracy claims rest on RadiusFPS producing identical samples to classical FPS via a conservative spherical-voxel bound plus coordinate-wise skip test. The abstract asserts preservation of the update rule and tie-breaking, but without an explicit proof or exhaustive equivalence check in the derivation, floating-point precision, degenerate configurations, or interaction between the two pruning stages could cause a different point to be selected in some iteration. Any such deviation would make the reported segmentation accuracy merely coincidental rather than guaranteed.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper proposes RadiusFPS, a framework for accelerating classical Farthest Point Sampling (FPS) on point clouds via spherical voxel pruning that is claimed to preserve the exact FPS update rule and tie-breaking policy. It introduces CPU and GPU (RadiusFPS-G) implementations with fused kernels, reporting up to 2.5x speedup over GPU FPS, roughly half the memory of QuickFPS, and comparable segmentation accuracy on S3DIS, ScanNet, and SemanticKITTI when used in downstream pipelines, including with learning-based samplers.","tokens_in":1917,"tokens_out":578,"duration_ms":33600,"significance":"If the claimed exact equivalence holds, the work would provide a practical, low-memory acceleration for a core operator in robotic perception pipelines handling high-rate LiDAR data. The explicit comparisons to QuickFPS on public benchmarks, memory measurements, and end-to-end timing with FastPoint are concrete strengths that would make the result useful for latency-constrained systems.","major_comments":[{"comment":"Abstract (pruning derivation paragraph): the central claim that spherical voxel indexing produces a conservative geometric bound safely pruning distance computations without discarding the true farthest point (under standard FPS update and tie-breaking) rests on an unshown geometric derivation. No explicit proof, exhaustive check, or handling of floating-point/degenerate cases is provided, making equivalence to classical FPS unverifiable and rendering the reported accuracy parity non-guaranteed.","section":"Abstract"},{"comment":"§4 (evaluation on S3DIS/ScanNet/SemanticKITTI): the speedup, memory, and accuracy claims are presented without error bars, multiple random seeds, or statistical tests. Given that any deviation in selected points from the unverified pruning stages would affect downstream segmentation, the absence of variability analysis undermines the reliability of the 'matches or exceeds QuickFPS' and 'comparable accuracy' statements.","section":"§4"}],"minor_comments":[{"comment":"The interaction between the spherical voxel pruning stage and the coordinate-wise point-skip test is described at a high level; a small example or pseudocode showing their combined effect on a single iteration would improve clarity without altering the central claim.","section":"§3"},{"comment":"Table captions and axis labels in the runtime/memory plots should explicitly state the number of points sampled (e.g., 1024 or 4096) and whether timings include voxel construction overhead.","section":null}],"recommendation":"major_revision","confidential_remarks":"The unshown derivation is the primary load-bearing gap; if the authors can supply a self-contained geometric argument or exhaustive equivalence test in revision, the paper would be substantially stronger. The work fits the journal's robotics/perception scope well."},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive feedback. We address each major comment below and will revise the manuscript to incorporate the suggested improvements.","responses":[{"response":"We agree that an explicit formal proof is needed for verifiability. Section 3 derives the conservative spherical voxel bound, but the revised manuscript will expand this with a complete geometric proof, including analysis of floating-point precision and degenerate cases (e.g., points on voxel boundaries or ties), to rigorously establish preservation of the exact FPS update rule and tie-breaking policy.","revision_made":"yes","referee_comment":"[Abstract] Abstract (pruning derivation paragraph): the central claim that spherical voxel indexing produces a conservative geometric bound safely pruning distance computations without discarding the true farthest point (under standard FPS update and tie-breaking) rests on an unshown geometric derivation. No explicit proof, exhaustive check, or handling of floating-point/degenerate cases is provided, making equivalence to classical FPS unverifiable and rendering the reported accuracy parity non-guaranteed."},{"response":"We acknowledge the lack of error bars, multi-seed analysis, and statistical tests in the current evaluation. The revised manuscript will report accuracy metrics averaged over multiple random seeds (minimum 5) with standard deviations for all datasets and include statistical tests to support the comparable accuracy claims.","revision_made":"yes","referee_comment":"[§4] §4 (evaluation on S3DIS/ScanNet/SemanticKITTI): the speedup, memory, and accuracy claims are presented without error bars, multiple random seeds, or statistical tests. Given that any deviation in selected points from the unverified pruning stages would affect downstream segmentation, the absence of variability analysis undermines the reliability of the 'matches or exceeds QuickFPS' and 'comparable accuracy' statements."}],"tokens_in":1495,"tokens_out":391,"duration_ms":34255,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The main thing here is a practical acceleration for farthest point sampling that keeps the original selection rule and tie-breaking intact. The new pieces are the spherical voxel indexing for a conservative distance bound, the extra coordinate-wise skip test, and the warp-level GPU kernel that fuses the steps to cut memory traffic.\n\nIt does the obvious things right for this kind of work: runs on both CPU and GPU, tests on S3DIS, ScanNet, and SemanticKITTI, shows speedups against plain GPU FPS and QuickFPS, reports roughly half the memory of QuickFPS, and keeps downstream segmentation accuracy comparable. The integration note with FastPoint is a small plus for end-to-end pipelines.\n\nThe soft spot is the pruning bound itself. The abstract says it is conservative and preserves semantics, but without the geometric derivation or an exhaustive check on edge cases (floating-point ties, degenerate point placements, interaction of the two pruning stages), it is hard to know whether the equivalence is guaranteed or just held on the reported runs. No error bars or repeated trials are mentioned, which is minor but worth noting for a performance claim.\n\nThis is for people who need faster FPS inside real-time LiDAR or point-cloud perception stacks and are willing to verify the exactness claim themselves. It is solid enough on the experimental side to deserve a serious referee rather than a desk reject, provided the bound proof is in the full text and holds up.","headline":"RadiusFPS claims exact FPS equivalence via spherical voxel pruning and a fused GPU kernel, delivering reported 2.5x speedups and lower memory on standard benchmarks, but the bound's conservativeness needs explicit checking.","tokens_in":2389,"tokens_out":375,"would_cite":false,"duration_ms":15775,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"Spherical voxel indexing derives a conservative bound that safely prunes redundant distance checks in farthest point sampling without altering the output.","keywords":["farthest point sampling","point cloud downsampling","spherical voxels","GPU acceleration","LiDAR processing","robotic perception","real-time sampling"],"falsifier":"Run both classical FPS and RadiusFPS on the same point cloud with identical initialization and tie-breaking; if any sampled point differs, the pruning bound is not conservative.","tokens_in":2706,"feed_emoji":"⚡","tokens_out":632,"duration_ms":24797,"temperature":0.7,"pith_summary":"The paper sets out to show that standard farthest point sampling can be made substantially faster on both CPUs and GPUs by indexing the input cloud in spherical voxels and using the resulting geometric bound to skip distance calculations that cannot affect the next selected point. A coordinate-wise skip test removes further redundant updates. The GPU kernel fuses voxel selection, pruning, and updates into coalesced operations. On indoor and outdoor LiDAR benchmarks the resulting sampler delivers the claimed speedups while producing identical sample sets to classical FPS and comparable downstream segmentation accuracy.","feed_headline":"Spherical voxels cut FPS distance work for 2.5x GPU speedup","feed_subtitle":"The bound keeps the exact classical output while halving memory use on indoor and outdoor LiDAR sets.","key_machinery":"Spherical voxel pruning: indexing the point cloud with spherical voxels to derive a conservative geometric bound that safely eliminates distance computations guaranteed not to contain the next farthest point.","core_discovery":"RadiusFPS indexes the point cloud with spherical voxels to obtain a conservative geometric bound that prunes redundant distance computations in each FPS iteration while preserving the exact update rule, initialization, and tie-breaking policy of standard FPS; a coordinate-wise point-skip test eliminates residual updates; the GPU variant fuses these steps into memory-coalesced kernels.","pith_inferences":["The same voxel-bound technique could be applied to other iterative farthest-point or diversity-sampling loops that rely on repeated distance queries.","Because the method preserves exact FPS output, it can serve as a drop-in replacement inside any pipeline that assumes the classical sampling distribution.","The coordinate-wise skip test may generalize to other coordinate-aligned data structures beyond spherical voxels."],"forward_implications":["RadiusFPS-G reaches up to 2.5 times the speed of GPU-based FPS on the evaluated indoor and outdoor datasets.","It matches or exceeds QuickFPS performance while using roughly half the GPU memory.","Segmentation accuracy remains comparable to standard FPS on S3DIS, ScanNet, and SemanticKITTI.","Pairing RadiusFPS-G with a learning-based sampler yields the fastest end-to-end inference among tested configurations."],"fun_headline_variants":["Voxels prune FPS for 2.5x GPU speedup with exact output","Spherical pruning matches classical FPS at 2.5x on GPU","Exact FPS via voxel bounds at 2.5x GPU speed and half memory","RadiusFPS uses voxels to speed exact FPS 2.5x on GPU"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The spherical voxel bound is guaranteed to be conservative, so that no point that would have been chosen by the standard FPS rule is ever pruned.","fun_headline_variants_meta":{"raw":{"variants":["Voxels prune FPS for 2.5x GPU speedup with exact output","Spherical pruning matches classical FPS at 2.5x on GPU","Exact FPS via voxel bounds at 2.5x GPU speed and half memory","RadiusFPS uses voxels to speed exact FPS 2.5x on GPU"]},"model":"grok-4.3","cost_usd":0.006922,"raw_usage":{"total_tokens":3249,"prompt_tokens":745,"num_sources_used":0,"completion_tokens":81,"cost_in_usd_ticks":69224500,"prompt_tokens_details":{"text_tokens":745,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2423,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":745,"tokens_out":81,"duration_ms":28696,"temperature":1.0,"reasoning_tokens":2423,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-06-28T00:59:13.916722+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Run both classical FPS and RadiusFPS on the same point cloud with identical initialization and tie-breaking; if any sampled point differs, the pruning bound is not conservative.","supporting_citations":[],"review_version":1}