Pith. sign in

REVIEW 4 cited by

3D FlowMatch Actor: Unified 3D Policy for Single- and Dual-Arm Manipulation

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2508.11002 v2 pith:UUAARPAR submitted 2025-08-14 cs.RO

Nikolaos Gkanatsios , Jiahe Xu , Matthew Bronars , Arsalan Mousavian , Tsung-Wei Ke , Katerina Fragkiadaki This is my paper · ORCID
classification cs.RO
keywords policyactionactordiffusion-basedflowflowmatchlearningmanipulation
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

We present 3D FlowMatch Actor (3DFA), a 3D policy architecture for robot manipulation that combines flow matching for trajectory prediction with 3D pretrained visual scene representations for learning from demonstration. 3DFA leverages 3D relative attention between action and visual tokens during action denoising, building on prior work in 3D diffusion-based single-arm policy learning. Through a combination of flow matching and targeted system-level and architectural optimizations, 3DFA achieves over 30x faster training and inference than previous 3D diffusion-based policies, without sacrificing performance. On the bimanual PerAct2 benchmark, it establishes a new state of the art, outperforming the next-best method by an absolute margin of 41.4%. In extensive real-world evaluations, it surpasses strong baselines with up to 1000x more parameters and significantly more pretraining. In unimanual settings, it sets a new state of the art on 74 RLBench tasks by directly predicting dense end-effector trajectories, eliminating the need for motion planning. Comprehensive ablation studies underscore the importance of our design choices for both policy effectiveness and efficiency.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Learning 3D Affordances for Blade Insertion in Cluttered Stowing

    cs.CV 2026-06 conditional novelty 6.5 of 10

    VulcanVoxel reconstructs blade occupancy with a 3D masked autoencoder, recovering multi-modal free-space affordances from unimodal warehouse stow data and raising top-5 coverage from 0.71 to 0.89.

  2. ARGUS: Aligning Robot Scene Geometry Under Shifting Views with Large 3D Vision Models

    cs.RO 2026-08 conditional novelty 6.0 of 10

    A preprocessing pipeline that reconstructs a 3D point cloud from RGB images and re-renders it from a fixed viewpoint improves viewpoint robustness and data efficiency for vision-based robot policies.

  3. High-Fidelity One-Step Generative Visuomotor Policy via Recursive Correction, Frequency Consistency, and Contrastive Flow Matching

    cs.RO 2026-07 conditional novelty 6.0 of 10

    One-step flow-matching visuomotor policy with recursive correction, dual-timestep spectral consistency, and contrastive mode separation matches or exceeds 10-step baselines at 1 NFE.

  4. ChronoFlow-Policy: Unifying Past-Current-Future Interaction Flow in Visuomotor Policy Learning

    cs.RO 2026-06 unverdicted novelty 6.0 of 10

    Co-training a diffusion visuomotor policy on unified past–current–future sparse 3D object–gripper keypoint flows improves long-horizon and non-Markovian manipulation over action-only and future-only baselines.

Pith tools