Weekly Top 10 · W46

2026-11-09 ~ 11-15 · Embodied AI

Generative · Personalization · AI Search · Infra · Embodied AI

  1. 1. From Video Frames to Robot Meta-Frames: A Unified Latent Interface for World Action Models · 📖 Deep dive

    When a video diffusion model is turned into a robot world action model, low-dimensional signals such as joint state, actions and value must enter a latent space that has only seen natural video…

    Appeared in: 11-12

  2. 2. Practice Makes Precise: At CoRL, NVIDIA and Partners Reveal How Robots Rehearse Delicate Work in Simulation

    Simulation has long been where robots learn to walk, and NVIDIA now wants to use it for fine work such as aligning gears and connecting cables, where the parts are millimeter-scale and training each task separately…

    Appeared in: 11-11

  3. 3. Real-Time Robot Execution with Masked Action Chunking

    Running the next model call while the robot executes the current action segment is meant to avoid pauses, yet done naively it lowers success. Earlier work blamed poor joins between consecutive segments…

    Appeared in: 11-13

  4. 4. Runway’s Praxis-1 world action model for robotics is now available

    Runway, a video generation company, has turned its video foundation model into a robot policy called Praxis-1, released with open weights under Apache 2.0 along with inference and fine-tuning code…

    Appeared in: 11-13

  5. 5. RoboColosseum: Evaluating Physical Intelligence at Scale In-The-Wild · 📖 Deep dive

    Ranking general-purpose robot policies is hard because real-robot experiments are difficult to reproduce: change the site or the object layout and the result changes. RoboColosseum tests five robot foundation models…

    Appeared in: 11-09

  6. 6. Kimodo-Policy: Learning Kinematic Action Models for Humanoid Loco-Manipulation

    When a humanoid walks while picking things up, its torso, arms and legs must coordinate continuously, yet existing vision-language-action and world action models inherit semantics and visual dynamics from pretraining…

    Appeared in: 11-11

  7. 7. Learn the Task, Not the Schedule: Reducing Spurious Cross-Arm Coupling in Bimanual Imitation Learning · 📖 Deep dive

    In two-armed robot demonstrations, which hand moves first is often just the operator's habit, yet imitation learning cannot separate the orderings a task requires from the ones that happened by chance…

    Appeared in: 11-11

  8. 8. APS-VLA: Action-Guided and Future-Aware Token Pruning for Efficient Vision-Language-Action Models · 📖 Deep dive

    A VLA processes hundreds of visual tokens at every step, which makes inference slow, yet the next action often depends on a small part of the image. This paper adds a small scorer in front of a frozen VLA and learns…

    Appeared in: 11-12

  9. 9. CIALLO: Learning Shared Intent Representations for Cross-Embodiment Manipulation

    Robots write actions in different ways, some as joint angles and some as end-effector displacements, so one robot's demonstrations are hard to reuse on another. This paper trains an action codec that aligns actions…

    Appeared in: 11-12

  10. 10. Zero-Shot VLA Adaptation with Retrieved Guidance

    Concatenating retrieved demonstrations into the model context is the standard RAG recipe for language models, but it does not carry over to robots. The authors retrieve demonstrations from real data such as DROID…

    Appeared in: 11-13

‹ Previous week · Next week ›