3DThinkVLA: Endowing Vision-Language-Action Models with Latent 3D Priors via 3D-Thinking-Guided Co-training

• URL: http://arxiv.org/abs/2606.04436v1
• Date: Wed, 03 Jun 2026 04:34:07 GMT
• Title: 3DThinkVLA: Endowing Vision-Language-Action Models with Latent 3D Priors via 3D-Thinking-Guided Co-training
• Abstract summary: 3D geometry perception and 3D spatial reasoning are distinct capabilities that can be disentangled and injected at different feature hierarchies.<n>We propose a 3D-thinking-guided co-training framework that enables vision-language-action (VLA) models to perform 3D spatial reasoning implicitly during action prediction.
• Score: 19.5808550016589
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We propose a 3D-thinking-guided co-training framework that enables vision-language-action (VLA) models to perform 3D spatial reasoning implicitly during action prediction. Our core insight is that 3D geometry perception and 3D spatial reasoning are distinct capabilities that can be disentangled and injected at different feature hierarchies. During training, three tightly coupled components work in concert primarily within the latent space: (1) To gain geometric priors, a latent 3D geometry perception module aligns intermediate visual features with a 3D foundation model, acquiring low-level geometric cues without architectural modifications to the VLM backbone. (2) Complementing this, an online 3D reasoning distillation module mitigates the prompt-induced reasoning gap via a shared reasoning anchor token. During 3D VLM co-training, this anchor is emitted as the first output token to robustly encode spatial priors. During VLA training, it serves as an input token inserted between the task and action instructions, transferring high-level spatial thinking from explicit teacher reasoning prompts to student action prompts without chain-of-thought text generation. (3) These disentangled geometric and reasoning features are then united by a spatially augmented action integration, which jointly injects them into the action-query tokens as hierarchical spatial conditions to prevent action shortcuts. At deployment, our method retains only its lightweight adapters to perform implicit 3D reasoning, discarding the 3D foundation model and the teacher branch used for supervision. Consequently, it operates purely on 2D images without 3D sensors, external models, or explicit text generation while preventing catastrophic forgetting of the pretrained VLM, achieving state-of-the-art performance on LIBERO, LIBERO-PLUS, SimplerEnv, and real-world manipulation tasks.

Related papers

• LAWM-3D: Learning 3D-Aware Latent Actions from Human Videos for Generalizable Robot World Models [64.11117186468533]
  World models enable agents to perform forward rollout and planning without real-world interaction.<n>Most existing LAMs rely on single-view inputs and operate primarily in 2D pixel space.<n>We propose LAWM-3D, which introduces three tightly coupled key designs for learning 3D-aware latent actions.
  arXiv  Detail & Related papers  (2026-08-06T07:47:01Z)
• Lift3D-VLA: Lifting VLA Models to 3D Geometry and Dynamics-Aware Manipulation [76.16065370700488]
  We introduce Lift3D-VLA, a unified VLA framework that equips models with explicit 3D point cloud reasoning and enables temporally coherent action generation.<n>We show that Lift3D-VLA achieves 10.8% and 11.1% higher mean success rates on MetaWorld and RLBench than the best-performing prior VLA methods.
  arXiv  Detail & Related papers  (2026-07-07T17:59:47Z)
• PointACT: Vision-Language-Action Models with Multi-Scale Point-Action Interaction [57.63073414949329]
  We propose PointACT, a dual-system 3D-aware VLA policy that integrates hierarchical 3D point cloud representations directly into the action decoding process.<n>PointACT employs a multi-scale point-action interaction mechanism with efficient bottleneck window self-attention, enabling evolving action tokens to densely attend to both local geometric detail and global scene structure.
  arXiv  Detail & Related papers  (2026-05-20T17:10:31Z)
• Proxy3D: Efficient 3D Representations for Vision-Language Models via Semantic Clustering and Alignment [50.02189698630855]
  We propose a Proxy3D method with compact yet comprehensive 3D proxy representations for the vision modality.<n>Our method achieves competitive or state-of-the-art performance in 3D visual question answering, visual grounding and general spatial intelligence benchmarks.
  arXiv  Detail & Related papers  (2026-05-08T17:50:47Z)
• GAP-MLLM: Geometry-Aligned Pre-training for Activating 3D Spatial Perception in Multimodal Large Language Models [70.61152292499737]
  We argue that this gap does not arise from insufficient geometric priors, but from a misalignment in the training paradigm.<n>Existing approaches typically resort to naive feature concatenation and optimize directly for downstream tasks without geometry-specific supervision.<n>We propose GAP-MLLM, a Geometry-Aligned Pre-training paradigm that explicitly activates structural perception before downstream adaptation.
  arXiv  Detail & Related papers  (2026-03-17T12:43:48Z)
• Abstract 3D Perception for Spatial Intelligence in Vision-Language Models [100.13033631690114]
  Vision-language models (VLMs) struggle with 3D-related tasks such as spatial cognition and physical understanding.<n>We introduce SandboxVLM, a framework that leverages abstract bounding boxes to encode geometric structure and physical kinematics for VLM.<n>Our approach consistently improves spatial intelligence, achieving an 8.3% gain on SAT Real compared with baseline methods.
  arXiv  Detail & Related papers  (2025-11-14T04:16:09Z)
• AffordBot: 3D Fine-grained Embodied Reasoning via Multimodal Large Language Models [20.05010202296243]
  We introduce Fine-grained 3D Embodied Reasoning, which requires an agent to predict, for each referenced affordance element in a 3D scene, its location, motion type, and motion axis.<n>We propose AffordBot, a novel framework that integrates Multimodal Large Language Models (MLLMs) with a tailored chain-of-thought (CoT) reasoning paradigm.<n>AffordBot achieves state-of-the-art performance, demonstrating strong generalization and physically grounded reasoning with only 3D point cloud input and MLLMs.
  arXiv  Detail & Related papers  (2025-11-13T06:43:00Z)
• Task-Aware 3D Affordance Segmentation via 2D Guidance and Geometric Refinement [12.260126771415019]
  We introduce Task-Aware 3D Scene-level Affordance segmentation (TASA)<n>TASA is a novel geometry-optimized framework that jointly leverages 2D semantic cues and 3D geometric reasoning in a coarse-to-fine manner.<n>To fully exploit 3D geometric information, a 3D affordance refinement module is proposed to integrate 2D semantic priors with local 3D geometry.
  arXiv  Detail & Related papers  (2025-11-12T13:36:37Z)
• From Spatial to Actions: Grounding Vision-Language-Action Model in Spatial Foundation Priors [54.84863164684646]
  Existing vision-language-action (VLA) models act in 3D real-world but are typically built on 2D encoders.<n>In this work, we introduce FALCON, a novel paradigm that injects rich 3D spatial tokens into the action head.
  arXiv  Detail & Related papers  (2025-10-20T11:26:45Z)
• Reg3D: Reconstructive Geometry Instruction Tuning for 3D Scene Understanding [6.7958985137291235]
  Reg3D is a novel Reconstructive Geometry Instruction Tuning framework that incorporates geometry-aware supervision directly into the training process.<n>Our key insight is that effective 3D understanding requires reconstructing underlying geometric structures rather than merely describing them.<n>Experiments on ScanQA, Scan2Cap, ScanRefer, and SQA3D demonstrate that Reg3D delivers substantial performance improvements.
  arXiv  Detail & Related papers  (2025-09-03T18:36:44Z)
• 3D-Aware Vision-Language Models Fine-Tuning with Geometric Distillation [17.294440057314812]
  Vision-Language Models (VLMs) have shown remarkable performance on diverse visual and linguistic tasks.<n>We propose Geometric Distillation, a framework that injects human-inspired geometric cues into pretrained VLMs.<n>Our method shapes representations to be geometry-aware while remaining compatible with natural image-text inputs.
  arXiv  Detail & Related papers  (2025-06-11T15:56:59Z)
• VL-SAT: Visual-Linguistic Semantics Assisted Training for 3D Semantic Scene Graph Prediction in Point Cloud [51.063494002003154]
  3D semantic scene graph (3DSSG) prediction in the point cloud is challenging since the 3D point cloud only captures geometric structures with limited semantics compared to 2D images. We propose Visual-Linguistic Semantics Assisted Training scheme that can significantly empower 3DSSG prediction models with discrimination about long-tailed and ambiguous semantic relations.
  arXiv  Detail & Related papers  (2023-03-25T09:14:18Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.