Bezen

Your daily digest

Research papers, made simple to understand

The AI and machine learning papers people are reading most, each one rewritten in plain words: the problem, the idea, what was found and why it matters.

Robotics & Physical AI9 papers, newest first

    • Robotics & Physical AI

    DONIC enables one policy for diverse bimanual dexterous manipulation from human data

    DONIC learns a single bimanual robot hand policy from a large multi-view dataset and public data, using object-centered rewards and Clip-Centered Advantages to achieve 86.79% success on unseen trajectories and 57.32% on unseen objects without fine-tuning.

    Paper: DONIC: Generalist Dexterous Bimanual Tracking Control from Human Demonstrations · Mingze Sun, Ziang Cao, Yuan Fei and 8 others

    129 upvotes on Hugging Face22 Oct 20266 min readRead the article

    • Robotics & Physical AI

    Ropedia W0 unifies human and robot loco-manipulation world modeling with large-scale egocentric data

    Ropedia W0 is a video diffusion transformer world model conditioned on camera poses and projected skeletons, trained on 10k hours of human and 3k hours of robot egocentric video, achieving state-of-the-art cross-embodiment prediction and enabling real-time interactive rollouts.

    Paper: Ropedia W0: Scaling Loco-Manipulation World Models across Embodiments with Human Experience · Tianqi Liu, Hanfei Zhao, Qifeng Liang and 12 others

    180 upvotes on Hugging Face21 Oct 20267 min readRead the article

    • Robotics & Physical AI

    EgoXperience enables joint 3D reconstruction of egocentric video components in a shared world frame

    EgoXperience introduces a 50-hour egocentric video dataset with aligned annotations and a unified model, EgoX, that jointly reconstructs camera pose, scene depth, hands, and objects in metric scale, reducing key errors by up to 68.9%.

    Paper: EgoXperience: Learning World-Grounded Egocentric Representations from Human Experience · Xinhao Ji, Hui En Pang, Hao Li and 12 others

    130 upvotes on Hugging Face20 Oct 20267 min readRead the article

    • AI Agents
    • Robotics & Physical AI

    HexaShell uses coding agents with RGB triangulation to improve robot manipulation on RoboDojo

    HexaShell lets a coding agent control a dual-arm robot via a thin command interface, evolving small tools from failures using RGB camera triangulation. It achieves 39.7% success on 42 RoboDojo tasks, surpassing learned policies, but struggles with tight insertion tasks requiring precise contact.

    Paper: HexaShell: A Thin Harness for Exploring the Limits of VLM Agents as Robot Coding Agents · Hongcheng Gao, Alex Zhou, Jay Zhu and 13 others

    285 upvotes on Hugging Face19 Oct 20266 min readRead the article

    • Robotics & Physical AI
    • AI Agents

    GPT-6-Astra shows broad robotic skills but faces real-world deployment challenges

    GPT-6-Astra achieves state-of-the-art results on 12 robotic benchmarks using in-context learning and extended interaction, but real-world tests reveal costly demonstrations, physical constraints, and the need for new benchmarks reflecting practical embodied intelligence.

    Paper: Has GPT-6-Astra Achieved Embodied Intelligence? · SenseNova, Joint Research Team, Zhongang Cai and 59 others

    240 upvotes on Hugging Face19 Oct 20267 min readRead the article

    • Multimodal & Vision
    • Robotics & Physical AI

    GroundingPI improves precise visual grounding for physical intelligence tasks

    GroundingPI is a 4B-parameter foundation model that generates quantized points and boxes for precise visual grounding, achieving state-of-the-art results on 34 benchmarks and improving robotic manipulation and autonomous driving performance.

    Paper: GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives · Qize Yu, Lianrui Fan, Boyu Chen and 23 others

    247 upvotes on Hugging Face13 Oct 20266 min readRead the article

    • Robotics & Physical AI

    Reflex enables humanoid robots to catch thrown boxes using egocentric RGB-D vision

    Reflex is a three-stage framework that trains a humanoid robot to catch thrown boxes by learning whole-body control, object dynamics inference, and visual grounding from egocentric RGB-D input, achieving 85.7% catch success in simulation and up to 65% on a real robot at 2.5 m.

    Paper: Reflex: Visually Grounded Reactive Humanoid Control · Taoyang Jia, Weikai Huang, Linxin Song and 6 others

    132 upvotes on Hugging Face12 Oct 20267 min readRead the article

    • Robotics & Physical AI
    • AI Agents

    MotorMind enables zero-shot robot manipulation using general vision-language models

    MotorMind uses a frozen general-purpose vision-language model to generate mid-level robot actions, with asynchronous monitoring and feedback, achieving 66.7% success on LIBERO-PRO and 95% on a real robot without task-specific training.

    Paper: MotorMind: Scaffolding General Vision Language Models for Zero-Shot Robot Manipulation · Bingxuan Li, Siqi Song, Yizhuo Wu and 3 others

    114 upvotes on Hugging Face10 Oct 20266 min readRead the article

    • Robotics & Physical AI

    Long-WAM improves robot control by scaling visual history with autoregressive video pretraining

    Long-WAM is a robot control framework that leverages long visual histories effectively by first autoregressively pretraining video prediction models, then adapting them for action generation, achieving up to 78.7% success on RoboCasa GR-1 with 19.2 seconds of context and real-time deployment on various hardware.

    Paper: Long-WAM: Scaling the Context of World-Action Models · Wei Huang, Bohan Zhang, Chenzhi Liu and 13 others

    90 upvotes on Hugging Face8 Oct 20266 min readRead the article