VLA language backbones show high redundancy on manipulation benchmarks, with half the LLM blocks removable and even two blocks sufficient to recover baseline performance after fine-tuning, unlike vision and action pathways.
hub Canonical reference
Spatial forcing: Implicit spatial representation alignment for vision- language-action model
Canonical reference. 71% of citing Pith papers cite this work as background.
34
Pith papers citing it
Background
71% of classified citations
hub tools
citation-role summary
background 5
baseline 1
method 1
citation-polarity summary
representative citing papers
VLMs exhibit consistent vertical-distance entanglement in embeddings from perspective bias in natural images, producing accuracy gaps that a new synthetic benchmark SpatialTunnel exposes as model-intrinsic.
VEGA improves spatial reasoning in VLA models for robotics by aligning visual encoder features with 3D-supervised DINOv2 representations via a temporary projector and cosine similarity loss.
Being-H0.7 adds future-aware latent reasoning to direct VLA policies via dual-branch alignment on latent queries, matching world-model benefits at VLA efficiency.
UniTacVLA builds a state-aware and dynamics-aware tactile prior via unified latent space, tactile chain-of-thought, and mixed real/predicted feedback controller to boost dexterous manipulation performance.
ImageWAM shows image editing models can replace video generation in world action models, delivering better performance with 6x lower FLOPs and 4x lower latency by using edit-derived KV caches as compact context.
GAM splits a geometric foundation model to enable language-conditioned future geometry prediction and action decoding for robot policies, claiming superior performance on manipulation benchmarks.
ReRe boosts open-source MLLMs on spatial reasoning benchmarks VSI-Bench and STI-Bench to rival proprietary SOTA by using a two-phase Reason then Re-reason process with Geometry-to-Video novel view synthesis.
FiberTune is a new fine-tuning objective that preserves action-fiber visual residuals in VLA policies, yielding performance gains on simulation and physical robot tasks.
AffordanceVLA proposes a VLA model with affordance-aware modules (Which2Act, Where2Act, How2Act) in a Mixture-of-Transformer trained in three stages to improve robotic manipulation.
A 3D-thinking-guided co-training method disentangles geometry perception and spatial reasoning to inject latent 3D priors into VLA models via adapters, achieving SOTA on manipulation benchmarks while running on 2D images only.
ERVLA trains on a 978k-trajectory embodied CoT corpus using reasoning as supervision with dropout, then predicts actions without CoT at test time, reaching 86.9% on LIBERO-Plus and 53.2% on VLABench.
History-conditioned short-horizon intent tokens stabilize chunked VLA policies under observation aliasing and raise success on AliasBench and standard robot benchmarks.
RoboMemArena is a new large-scale robotic memory benchmark with real-world tasks, and PrediMem is a dual VLA system that outperforms baselines by managing memory buffers with predictive coding.
A unified comparison of latent action supervision strategies for VLA models reveals task-specific benefits, with image-based approaches aiding reasoning and generalization, action-based aiding motor control, and discrete tokens proving most effective.
X-WAM unifies robotic action execution and 4D world synthesis by adapting video diffusion priors with a lightweight depth branch and asynchronous noise sampling, achieving 79-91% success on robot benchmarks.
DeLock mitigates lock-in in low-data VLA post-training via visual grounding preservation and test-time contrastive prompt guidance, outperforming baselines across eight evaluations while matching data-heavy generalist policies.
Pretraining a vision-language model to output discrete 3D pose tokens on large non-robotic data, before training a robot action head, improves downstream manipulation success and data efficiency.
Robo3R predicts accurate metric-scale 3D scene geometry from RGB images and robot states for improved robotic manipulation performance.
LingBot-VLA is a VLA foundation model trained on massive real robot data that shows superior generalization across tasks and platforms with fast training throughput.
HiF-VLA improves long-horizon robotic manipulation by encoding past motion as hindsight priors and anticipating future motion through foresight reasoning inside a VLA framework.
Projecting robot end-effector state onto image feature maps and sampling co-located visual tokens improves manipulation policy success by 4-10% across 67 tasks.
A two-stage framework pretrains an action module with temporal motion priors from unconditioned trajectories using flow-matching, then transfers it to VLA training via decoder reuse and distillation, yielding better performance on cross-embodiment tasks.
SVP-IL decouples semantic reasoning from geometric grounding in vision-language-action models by injecting zero-shot spatial masks as explicit prompts into a continuous action generator, yielding higher success on ambiguous manipulation tasks with limited demonstrations.
citing papers explorer
-
Drop-Then-Recovery: How Redundant Are Vision-Language-Action Models?
VLA language backbones show high redundancy on manipulation benchmarks, with half the LLM blocks removable and even two blocks sufficient to recover baseline performance after fine-tuning, unlike vision and action pathways.
-
Why Far Looks Up: Probing Spatial Representation in Vision-Language Models
VLMs exhibit consistent vertical-distance entanglement in embeddings from perspective bias in natural images, producing accuracy gaps that a new synthetic benchmark SpatialTunnel exposes as model-intrinsic.
-
VEGA: Visual Encoder Grounding Alignment for Spatially-Aware Vision-Language-Action Models
VEGA improves spatial reasoning in VLA models for robotics by aligning visual encoder features with 3D-supervised DINOv2 representations via a temporary projector and cosine similarity loss.
-
Being-H0.7: A Latent World-Action Model from Egocentric Videos
Being-H0.7 adds future-aware latent reasoning to direct VLA policies via dual-branch alignment on latent queries, matching world-model benefits at VLA efficiency.
-
UniTacVLA: Unified Tactile Understanding and Prediction in Vision Language Action Models
UniTacVLA builds a state-aware and dynamics-aware tactile prior via unified latent space, tactile chain-of-thought, and mixed real/predicted feedback controller to boost dexterous manipulation performance.
-
ImageWAM: Do World Action Models Really Need Video Generation, or Just Image Editing?
ImageWAM shows image editing models can replace video generation in world action models, delivering better performance with 6x lower FLOPs and 4x lower latency by using edit-derived KV caches as compact context.
-
Geometric Action Model for Robot Policy Learning
GAM splits a geometric foundation model to enable language-conditioned future geometry prediction and action decoding for robot policies, claiming superior performance on manipulation benchmarks.
-
Reason, Then Re-reason: Cross-view Revisiting Improves Spatial Reasoning
ReRe boosts open-source MLLMs on spatial reasoning benchmarks VSI-Bench and STI-Bench to rival proprietary SOTA by using a two-phase Reason then Re-reason process with Geometry-to-Video novel view synthesis.
-
FiberTune: Preserving Action-Fiber Visual Residuals in Vision-Language-Action Fine-Tuning
FiberTune is a new fine-tuning objective that preserves action-fiber visual residuals in VLA policies, yielding performance gains on simulation and physical robot tasks.
-
AffordanceVLA: A Vision-Language-Action Model Empowering Action Generation through Affordance-Aware Understanding
AffordanceVLA proposes a VLA model with affordance-aware modules (Which2Act, Where2Act, How2Act) in a Mixture-of-Transformer trained in three stages to improve robotic manipulation.
-
3DThinkVLA: Endowing Vision-Language-Action Models with Latent 3D Priors via 3D-Thinking-Guided Co-training
A 3D-thinking-guided co-training method disentangles geometry perception and spatial reasoning to inject latent 3D priors into VLA models via adapters, achieving SOTA on manipulation benchmarks while running on 2D images only.
-
Revisiting Embodied Chain-of-Thought for Generalizable Robot Manipulation
ERVLA trains on a 978k-trajectory embodied CoT corpus using reasoning as supervision with dropout, then predicts actions without CoT at test time, reaching 86.9% on LIBERO-Plus and 53.2% on VLABench.
-
IntentVLA: Short-Horizon Intent Modeling for Aliased Robot Manipulation
History-conditioned short-horizon intent tokens stabilize chunked VLA policies under observation aliasing and raise success on AliasBench and standard robot benchmarks.
-
RoboMemArena: A Comprehensive and Challenging Robotic Memory Benchmark
RoboMemArena is a new large-scale robotic memory benchmark with real-world tasks, and PrediMem is a dual VLA system that outperforms baselines by managing memory buffers with predictive coding.
-
From Pixels to Tokens: A Systematic Study of Latent Action Supervision for Vision-Language-Action Models
A unified comparison of latent action supervision strategies for VLA models reveals task-specific benefits, with image-based approaches aiding reasoning and generalization, action-based aiding motor control, and discrete tokens proving most effective.
-
Unified 4D World Action Modeling from Video Priors with Asynchronous Denoising
X-WAM unifies robotic action execution and 4D world synthesis by adapting video diffusion priors with a lightweight depth branch and asynchronous noise sampling, achieving 79-91% success on robot benchmarks.
-
Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Training
DeLock mitigates lock-in in low-data VLA post-training via visual grounding preservation and test-time contrastive prompt guidance, outperforming baselines across eight evaluations while matching data-heavy generalist policies.
-
PoseVLA: Universal Pose Pretraining for Generalizable Vision-Language-Action Policies
Pretraining a vision-language model to output discrete 3D pose tokens on large non-robotic data, before training a robot action head, improves downstream manipulation success and data efficiency.
-
Robo3R: Enhancing Robotic Manipulation with Accurate Feed-Forward 3D Reconstruction
Robo3R predicts accurate metric-scale 3D scene geometry from RGB images and robot states for improved robotic manipulation performance.
-
A Pragmatic VLA Foundation Model
LingBot-VLA is a VLA foundation model trained on massive real robot data that shows superior generalization across tasks and platforms with fast training throughput.
-
HiF-VLA: Hindsight, Insight and Foresight through Motion Representation for Vision-Language-Action Models
HiF-VLA improves long-horizon robotic manipulation by encoding past motion as hindsight priors and anticipating future motion through foresight reasoning inside a VLA framework.
-
GeoProp: Grounding Robot State in Vision for Generalist Manipulation
Projecting robot end-effector state onto image feature maps and sampling co-located visual tokens improves manipulation policy success by 4-10% across 67 tasks.
-
Learning Action Priors for Cross-embodiment Robot Manipulation
A two-stage framework pretrains an action module with temporal motion priors from unconditioned trajectories using flow-matching, then transfers it to VLA training via decoder reuse and distillation, yielding better performance on cross-embodiment tasks.
-
Decoupling Semantics and Geometric Grounding: Spatial Visual Prompts for Language-Conditioned Imitation Learning
SVP-IL decouples semantic reasoning from geometric grounding in vision-language-action models by injecting zero-shot spatial masks as explicit prompts into a continuous action generator, yielding higher success on ambiguous manipulation tasks with limited demonstrations.
-
SpatialSV: Internalizing Interpretable 3D Spatial Awareness in MLLMs via Task-Oriented Visual Supervision
SpatialSV trains MLLMs to lift 2D visual features into explicit 3D representations via task-oriented supervision for interpretable spatial awareness.
-
GEM: Generative Supervision Helps Embodied Intelligence
GEM adds generative depth supervision to VLM pre-training and reports improved results on embodied benchmarks plus real-world robot execution.
-
PokeVLA: Empowering Pocket-Sized Vision-Language-Action Model with Comprehensive World Knowledge Guidance
PokeVLA is a lightweight VLA model pre-trained on 2.4M samples for spatial grounding and reasoning, then adapted via multi-view semantics and geometry alignment to achieve state-of-the-art robot manipulation performance.
-
R3D: Revisiting 3D Policy Learning
A transformer 3D encoder plus diffusion decoder architecture, with 3D-specific augmentations, outperforms prior 3D policy methods on manipulation benchmarks by improving training stability.
-
Unleashing More Actions via Action Compositional Training for VLA Models
ACT-VLA synthesizes novel demonstrations from existing VLA tasks via latent representations to reduce overfitting and improve generalization on manipulation tasks in simulation.
-
Evo-Depth: A Lightweight Depth-Enhanced Vision-Language-Action Model
Evo-Depth is a compact VLA model using a lightweight implicit depth encoder from RGB views plus progressive alignment to boost manipulation performance without added hardware.
-
OmniVLA-RL: A Vision-Language-Action Model with Spatial Understanding and Online RL
OmniVLA-RL uses a mix-of-transformers architecture and flow-matching reformulated as SDE with group segmented policy optimization to surpass prior VLA models on LIBERO benchmarks.
-
Understanding the Impact of Geometric Foundation Models on Vision-Language-Action Models
The paper quantifies the geometric gap in current VLAs via linear probing and compares three architectures for injecting geometry from GFMs while analyzing impacts of data, cameras, and reconstruction quality.
- EvoScene-VLA: Evolving Scene Beliefs Inside the Action Decoder for Chunked Robot Control
- SpatialVAM:Spatial-Aware Multi-View Video Diffusion as a Data-Efficient Robot Policy