SmolVLA
PulseAugur coverage of SmolVLA — every cluster mentioning SmolVLA across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
New VLA Model Improves Robot Manipulation Robustness and Interpretability
Researchers have developed CoBoVLA, a novel 250M-parameter Vision-Language-Action (VLA) model designed to enhance robustness and interpretability in robot manipulation. Unlike previous models that jointly process langua…
-
New research explores vision-language model advancements and safety
Multiple research papers published on arXiv explore advancements in vision-language models (VLMs). ViLFlow introduces a parallel generation approach for improved VQA and image captioning. MoMoVL focuses on modality-awar…
-
逐际动力 FluxVLA Engine V2 streamlines embodied AI development
逐际动力 (LimX Dynamics) has released the second version of its open-source engineering foundation, FluxVLA Engine, designed to streamline the development and deployment of embodied AI. This update enhances capabilities by …
-
New Backtrack method improves VLA model reliability in robotics
Researchers have developed a new method called Backtrack to improve the reliability of Vision-Language-Action (VLA) models in robotic manipulation. This lightweight mechanism detects and recovers from failures without r…
-
SmolVLA-MoE enhances robotic manipulation with sparse Mixture-of-Experts
Researchers have introduced SmolVLA-MoE, a novel framework that enhances vision-language-action (VLA) models for robotic manipulation by incorporating a sparse Mixture-of-Experts (MoE) approach. This method replaces den…
-
New method estimates robot policy uncertainty from single model
Researchers have developed a novel method for estimating epistemic uncertainty in Vision-Language-Action (VLA) policies, which are crucial for robots to identify knowledge gaps and request assistance. This new approach,…
-
Study reveals VLA models' motor programs tied to absolute coordinates, ignoring scene changes
Researchers have conducted a mechanistic interpretability study on vision-language-action (VLA) models, revealing that motor programs are often tied to absolute coordinates and disregard changes in the visual scene. Thi…
-
VLA-Rolling framework reuses discarded actions for faster, more accurate VLA models
Researchers have developed VLA-Rolling, a new framework that improves the efficiency and control of vision-language-action (VLA) models. This method reuses discarded action tails from previous planning cycles, rather th…
-
New framework WareFly-VLA enables language-guided UAV control in warehouses
Researchers have introduced WareFly-VLA, a new framework and dataset designed for vision-language-action (VLA) models to control unmanned aerial vehicles (UAVs) in smart warehouses. The framework includes a dataset of h…
-
New research explores VLA model efficiency and latency trade-offs · 2 sources tracked
Two new research papers explore the efficiency and performance of Vision-Language-Action (VLA) models. The first paper analyzes SmolVLA, demonstrating how deployment optimizations like ONNX can significantly reduce late…
-
New GUARD method detects failures in diffusion-based VLA policies
Researchers have developed GUARD, a novel method for detecting failures in diffusion-based vision-language-action (VLA) policies. GUARD operates at test time by analyzing the influence of specific tokens within the mode…
-
New framework enhances lightweight models for robotic control
Researchers have developed XS-VLA, a novel two-stage framework designed to enhance robotic control using lightweight vision-language models. The framework addresses the limitations of large models in real-time applicati…
-
CorridorVLA introduces explicit spatial constraints for generative action models
Researchers have introduced CorridorVLA, a novel approach for Vision-Language-Action (VLA) models that explicitly incorporates spatial constraints. Unlike previous methods that implicitly embed spatial guidance, Corrido…
-
Data poisoning attacks found practical and stealthy on open-source robotics models
Researchers have demonstrated that data poisoning attacks on open-source robotics models are practical and stealthy. By injecting a small number of poisoned samples into training data, a backdoor can be embedded into vi…
-
New AI system automates biological lab tasks with vision and language
Researchers have developed BioProVLA-Agent, an affordable system for automating biological laboratory tasks using Vision-Language-Action (VLA) models. This system integrates protocol parsing, visual state verification, …