FastWAM
PulseAugur coverage of FastWAM — every cluster mentioning FastWAM across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
New SteerTrace attack manipulates World Action Models with minimal perturbation
Researchers have developed SteerTrace, a novel attack method targeting World Action Models (WAMs) that can manipulate their predicted future actions. This attack requires only a minimal perturbation, shared across frame…
-
Just Action Model unifies VLA and WAM approaches for robot control
Researchers have introduced the Just Action Model (JAM), a novel framework for robot control that decouples key components of policy training. JAM separates backbone initialization, future observation objectives, and in…
-
UniWALM model unifies video, action, and language for robot control
Researchers have introduced UniWALM, a novel model that unifies video, action, and language processing within a single transformer backbone. This approach aims to enhance robot control by allowing modalities to co-denoi…
-
New adversarial attack TAPDreamer cripples robotic world action models
Researchers have developed TAPDreamer, a novel adversarial attack targeting world action models used in robotics. This attack generates fixed local perturbations that can be applied to camera inputs, significantly degra…
-
New World Action Models Enhance Generalization with Causal Semantics and Multi-Modal Prediction
Researchers have developed new world action models (WAMs) that improve generalization capabilities under visual distribution shifts. The first model, CSWAM, integrates a causal semantic expert built on V-JEPA 2.1 to bet…
-
New DECOWAM model enhances robot manipulation with decoupled actions
Researchers have developed DECOWAM, a new world-action model designed for legged mobile manipulation. This model distinguishes between camera ego-motion and base/arm actions, improving future video and action prediction…
-
New SoftVTBench dataset evaluates physical interaction in deformable object manipulation
Researchers have introduced SoftVTBench, a new dataset and benchmark designed to evaluate the quality of physical interaction in deformable object manipulation. This dataset pairs visual observations with tactile data, …
-
LeRobot v0.6.0 adds world models, new VLAs, and improved datasets
LeRobot v0.6.0 has been released, introducing significant advancements in robotics AI. The update features new world model policies like VLA-JEPA, FastWAM, and LingBot-VA, which enable robots to "imagine" future scenari…
-
AttenA+ framework boosts robotic foundation models by prioritizing critical actions
Researchers have introduced AttenA+, a novel framework designed to improve the performance of robotic foundation models. This architecture-agnostic approach addresses the issue of temporal homogeneity in training by rew…
-
AttenA+ framework boosts robotic foundation models with velocity-aware training
Researchers have developed AttenA+, a new framework designed to improve robotic foundation models by addressing action inequality during training. The framework prioritizes kinematically critical segments of robot traje…