Pith. sign in

REVIEW 13 cited by

PointVLA: Injecting the 3D World into Vision-Language-Action Models

Not yet reviewed by Pith; the record is open.

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 2503.07511 v1 pith:V76JANKC submitted 2025-03-10 cs.RO cs.CVcs.LG

classification cs.ROcs.CVcs.LG
keywords pointvlataskscloudmodelsobjectspointacrossaction
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Vision-Language-Action (VLA) models excel at robotic tasks by leveraging large-scale 2D vision-language pretraining, but their reliance on RGB images limits spatial reasoning critical for real-world interaction. Retraining these models with 3D data is computationally prohibitive, while discarding existing 2D datasets wastes valuable resources. To bridge this gap, we propose PointVLA, a framework that enhances pre-trained VLAs with point cloud inputs without requiring retraining. Our method freezes the vanilla action expert and injects 3D features via a lightweight modular block. To identify the most effective way of integrating point cloud representations, we conduct a skip-block analysis to pinpoint less useful blocks in the vanilla action expert, ensuring that 3D features are injected only into these blocks--minimizing disruption to pre-trained representations. Extensive experiments demonstrate that PointVLA outperforms state-of-the-art 2D imitation learning methods, such as OpenVLA, Diffusion Policy and DexVLA, across both simulated and real-world robotic tasks. Specifically, we highlight several key advantages of PointVLA enabled by point cloud integration: (1) Few-shot multi-tasking, where PointVLA successfully performs four different tasks using only 20 demonstrations each; (2) Real-vs-photo discrimination, where PointVLA distinguishes real objects from their images, leveraging 3D world knowledge to improve safety and reliability; (3) Height adaptability, Unlike conventional 2D imitation learning methods, PointVLA enables robots to adapt to objects at varying table height that unseen in train data. Furthermore, PointVLA achieves strong performance in long-horizon tasks, such as picking and packing objects from a moving conveyor belt, showcasing its ability to generalize across complex, dynamic environments.

Discussion (0). Sign in to comment.

Forward citations

Cited by 13 Pith papers

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

  1. Look Where It Matters: Adaptive Visual Refinement for Vision-Language-Action Models

    cs.RO 2026-08 conditional novelty 6.0 of 10

    Inserting register tokens into a VLA encoder plus uncertainty-gated, attention-guided cropping raises π0's success from 94.2% to 98.4% on LIBERO and 46.5% to 69.0% on a real-world benchmark, at 1.4–1.6× compute.

  2. VLA Knows Its Limits: Adaptive Execution Horizons for Robot Policies

    cs.RO 2026-02 conditional novelty 6.0 of 10

    AutoHorizon estimates per-chunk execution horizons in flow-based VLA policies from action self-attention plateau positions, beating fixed-horizon tuning in most tested benchmarks.

  3. Learning to Feel the Future: DreamTacVLA for Contact-Rich Manipulation

    cs.RO 2025-12 unverdicted novelty 6.0 of 10

    DreamTacVLA grounds VLA models in contact physics by aligning multi-scale vision-tactile inputs and predicting future tactile states, reaching up to 95% success on contact-rich tasks.

  4. StereoVLA: Enhancing Vision-Language-Action Models with Stereo Vision

    cs.RO 2025-12 conditional novelty 6.0 of 10

    A vision-language-action model that fuses stereo-derived geometric features with semantic features improves real-world grasping success and camera-pose robustness over single-view baselines.

  5. GeoVLA: Empowering 3D Representations in Vision-Language-Action Models

    cs.RO 2025-08 conditional novelty 6.0 of 10

    A robot policy that combines 2D vision-language features with a point-cloud encoder and a mixture-of-experts diffusion action head reports SOTA manipulation success in simulation and robust real-world behavior under h...

  6. AimBot: A Simple Auxiliary Visual Cue to Enhance Spatial Awareness of Visuomotor Policies

    cs.RO 2025-08 conditional novelty 6.0 of 10

    Overlaying end-effector-derived shooting lines and reticles on RGB images consistently raises success rates of visuomotor policies, especially on long-horizon manipulation tasks.

  7. T-Rex: Task-Adaptive Spatial Representation Extraction for Robotic Manipulation with Vision-Language Models

    cs.RO 2025-06 conditional novelty 6.0 of 10

    A zero-training framework that adaptively selects spatial representation extractors per object and per task stage improves real-world robot manipulation success and efficiency over fixed-representation baselines.

  8. AntiGrounding: Lifting Robotic Actions into VLM Representation Space for Decision Making

    cs.RO 2025-06 conditional novelty 6.0 of 10

    AntiGrounding lifts candidate robot trajectories into the VLM's visual space via multi-view rendering and structured VQA, and reports 57.5% average success across eight manipulation tasks, beating three intermediate-r...

  9. ChatVLA-2: Vision-Language-Action Model with Open-World Embodied Reasoning from Pretrained Knowledge

    cs.RO 2025-05 conditional novelty 6.0 of 10

    ChatVLA-2 uses dynamic mixture-of-experts and a two-stage training recipe to let a vision-language-action model retain pretrained reasoning while following robot instructions.

  10. VistaVLA: Geometry- and Semantic-Aware 3D Gaussian-Grounded VLA for Robotic Manipulation

    cs.RO 2026-07 conditional novelty 5.5 of 10

    VistaVLA lifts multi-view vision-language features into 3D Gaussians, compresses them 99% via Merge-then-Query, and improves real-robot manipulation success by ~23% over baselines.

  11. QDepth-VLA: Quantized Depth Prediction as Auxiliary Supervision for Vision-Language-Action Models

    cs.CV 2025-10 conditional novelty 5.0 of 10

    Adding an auxiliary quantized-depth-token prediction task to a VLA policy improves manipulation success rates on LIBERO, Simpler, and real-robot pick-and-place tasks versus the open-pi-zero baseline.

  12. Large Model Empowered Embodied AI: A Survey on Decision-Making and Embodied Learning

    cs.RO 2025-08 reject novelty 4.0 of 10

    A review that categorizes large-model-empowered embodied AI into hierarchical and end-to-end decision-making, imitation and reinforcement learning, and world models.

  13. A Survey on Vision-Language-Action Models for Autonomous Driving

    cs.CV 2025-06 conditional novelty 4.0 of 10

    A survey organizes vision-language-action models for autonomous driving into four stages, compares over 20 systems, and catalogs datasets, benchmarks, and open challenges.

Pith tools