A prompt-only attack called command-preserving trajectory redirection can steer VLA robot behavior to attacker-chosen physical outcomes while the text still appears to match the intended task.
hub
Vla-0: Building state-of-the-art vlas with zero modification
20 Pith papers cite this work. Polarity classification is still indexing.
20
Pith papers citing it
hub tools
citation-role summary
background 3
baseline 1
citation-polarity summary
representative citing papers
Pion modifies Muon's Newton-Schulz iterations into a controllable high-pass filter that anchors dominant singular values at 1 while suppressing noisy tails, outperforming Muon and AdamW in VLA and RLVR regimes.
VLATIM benchmark reveals large VLMs excel at high-level planning in physics puzzles but struggle with precise visual grounding and mouse control, so they lack human-like problem-solving capabilities.
Reducing visual input to one token per frame in VLA world models maintains or improves long-horizon performance on MetaWorld, LIBERO, and real-robot tasks.
A multimodal transformer generates and caches interleaved text-image traces to guide closed-loop actions, achieving 92.4% success on LIBERO-Long and 95.5% average on LIBERO.
UniLACT improves VLA models by adding depth-aware unified latent action pretraining that outperforms RGB-only baselines on seen and unseen manipulation tasks.
A training-free cache of intermediate actions, retrieved by random-projected VLM embeddings, cuts flow-based VLA action-head latency up to 40× in tests while keeping success rates near base-model levels.
UniviewVLA generates multiview future views from two cameras via world modeling, plus token compression and view selection, to boost occlusion handling in robot manipulation while matching standard benchmark performance.
w²VLA restructures VLA information flow to decouple declarative semantics from procedural skills, enabling zero-shot transfer to novel objects.
OneVLA is a unified VLA model using a shared action head and multi-stage progressive training with CoT fine-tuning that reports state-of-the-art results on both navigation and manipulation in simulation and real-world settings.
ALAM introduces algebraic consistency regularization on latent action transitions from videos, raising VLA success rates from 47.9% to 85.0% on MetaWorld MT50 and 94.1% to 98.1% on LIBERO.
Sparse action-aligned spatial anchors impose an explicit tolerance corridor that improves flow-matching VLA success rates by 4–12 points on LIBERO and LIBERO-Plus.
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.
VLANeXt distills 12 design insights from a unified VLA study into a model that outperforms prior methods on LIBERO benchmarks while releasing code for further exploration.
EyeVLA transfers open-world VLM understanding to a PTZ camera control policy via hierarchical action tokens and GRPO reinforcement learning, reaching 96% task completion on 50 real scenes with only 500 training samples.
AsyncVLA adds asynchronous flow matching and a confidence rater to VLA models so they can generate actions on flexible schedules and selectively refine low-confidence tokens before execution.
Sharpness-aware minimization during VLA finetuning preserves instruction following and yields over 60% gains across simulation and real-world tasks.
TBD-VLA partitions action sequences into temporal blocks, performs masked discrete diffusion within blocks, and autoregressive generation across blocks to unify parallel decoding with temporal coherence in discrete VLA models.
The World-Value-Action model enables implicit planning for VLA systems by performing inference over a learned latent representation of high-value future trajectories instead of direct action prediction.
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.
citing papers explorer
-
Trajectory-Level Redirection Attacks on Vision-Language-Action Models
A prompt-only attack called command-preserving trajectory redirection can steer VLA robot behavior to attacker-chosen physical outcomes while the text still appears to match the intended task.
-
Rethinking Muon Beyond Pretraining: Spectral Failures and High-Pass Remedies for VLA and RLVR
Pion modifies Muon's Newton-Schulz iterations into a controllable high-pass filter that anchors dominant singular values at 1 while suppressing noisy tails, outperforming Muon and AdamW in VLA and RLVR regimes.
-
Do Vision-Language-Models show human-like logical problem-solving capability in point and click puzzle games?
VLATIM benchmark reveals large VLMs excel at high-level planning in physics puzzles but struggle with precise visual grounding and mouse control, so they lack human-like problem-solving capabilities.
-
One Token Per Frame: Reconsidering Visual Bandwidth in World Models for VLA Policy
Reducing visual input to one token per frame in VLA world models maintains or improves long-horizon performance on MetaWorld, LIBERO, and real-robot tasks.
-
Thinking in Text and Images: Interleaved Vision--Language Reasoning Traces for Long-Horizon Robot Manipulation
A multimodal transformer generates and caches interleaved text-image traces to guide closed-loop actions, achieving 92.4% success on LIBERO-Long and 95.5% average on LIBERO.
-
UniLACT: Depth-Aware RGB Latent Action Learning for Vision-Language-Action Models
UniLACT improves VLA models by adding depth-aware unified latent action pretraining that outperforms RGB-only baselines on seen and unseen manipulation tasks.
-
ActionCache: Training-Free Acceleration for Vision-Language-Action Models with Action Caching and Refinement
A training-free cache of intermediate actions, retrieved by random-projected VLM embeddings, cuts flow-based VLA action-head latency up to 40× in tests while keeping success rates near base-model levels.
-
UniviewVLA: A Unified Multiview Vision-Language-Action Model with World Modeling
UniviewVLA generates multiview future views from two cameras via world modeling, plus token compression and view selection, to boost occlusion handling in robot manipulation while matching standard benchmark performance.
-
Decoupling the Declarative from the Procedural in Vision-Language-Action Models
w²VLA restructures VLA information flow to decouple declarative semantics from procedural skills, enabling zero-shot transfer to novel objects.
-
OneVLA: A Unified Framework for Embodied Tasks
OneVLA is a unified VLA model using a shared action head and multi-stage progressive training with CoT fine-tuning that reports state-of-the-art results on both navigation and manipulation in simulation and real-world settings.
-
ALAM: Algebraically Consistent Latent Action Model for Vision-Language-Action Models
ALAM introduces algebraic consistency regularization on latent action transitions from videos, raising VLA success rates from 47.9% to 85.0% on MetaWorld MT50 and 94.1% to 98.1% on LIBERO.
-
CorridorVLA: Explicit Spatial Constraints for Generative Action Heads via Sparse Anchors
Sparse action-aligned spatial anchors impose an explicit tolerance corridor that improves flow-matching VLA success rates by 4–12 points on LIBERO and LIBERO-Plus.
-
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.
-
VLANeXt: Recipes for Building Strong VLA Models
VLANeXt distills 12 design insights from a unified VLA study into a model that outperforms prior methods on LIBERO benchmarks while releasing code for further exploration.
-
Look, Zoom, Understand: The Robotic Eyeball for Embodied Perception
EyeVLA transfers open-world VLM understanding to a PTZ camera control policy via hierarchical action tokens and GRPO reinforcement learning, reaching 96% task completion on 50 real scenes with only 500 training samples.
-
AsyncVLA: Asynchronous Flow Matching for Vision-Language-Action Models
AsyncVLA adds asynchronous flow matching and a confidence rater to VLA models so they can generate actions on flexible schedules and selectively refine low-confidence tokens before execution.
-
Flatness Preserves Instruction Following in Vision-Language-Action Models
Sharpness-aware minimization during VLA finetuning preserves instruction following and yields over 60% gains across simulation and real-world tasks.
-
TBD-VLA: Temporal Block Diffusion Vision Language Action Model
TBD-VLA partitions action sequences into temporal blocks, performs masked discrete diffusion within blocks, and autoregressive generation across blocks to unify parallel decoding with temporal coherence in discrete VLA models.
-
World-Value-Action Model: Implicit Planning for Vision-Language-Action Systems
The World-Value-Action model enables implicit planning for VLA systems by performing inference over a learned latent representation of high-value future trajectories instead of direct action prediction.
-
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.