<p><span class="legendSpanClass"><span class="xn-location">BEIJING</span></span>, <span class="legendSpanClass"><span class="xn-chron">Nov. 16, 2025</span></span> /PRNewswire/ -- <b>Galbot</b> is proud to announce two major technological advancements in <b>embodied intelligence</b>: <b>DexNDM</b>, a neural dynamics model designed to revolutionize <b>robotic dexterous manipulation</b>, and <b>NavFoM</b>, the world's first <b>cross-embodiment, cross-task navigation foundation model</b>. Developed in collaboration with <b>Tsinghua University</b>, <b>Peking University</b>, <b>University of <span class="xn-location">Adelaide</span></b>, and <b><span class="xn-location">Zhejiang</span> University</b>, these innovations mark a significant leap forward in robotics, enabling robots to autonomously perform <b>complex tasks</b> across diverse environments.</p>  <div class="PRN_ImbeddedAssetReference" id="DivAssetPlaceHolder1" align="center">   <p title="Watch DexNDM empower robots with the precision to perform high-accuracy, in-hand rotations across a wide range of object types. From delicate microelectronics to complex geometries, DexNDM redefines robotic dexterity with its groundbreaking neural dynamics model."><iframe width="560" height="315" src="https://www.youtube.com/embed/7R9n1u2_MHs" frameborder="0">&amp;amp;amp;amp;amp;nbsp;</iframe><br /><span>Watch DexNDM empower robots with the precision to perform high-accuracy, in-hand rotations across a wide range of object types. From delicate microelectronics to complex geometries, DexNDM redefines robotic dexterity with its groundbreaking neural dynamics model.</span></p>  </div>  <p><b>DexNDM: A Game-Changer in Dexterous Manipulation</b></p>  <p><b>DexNDM</b>&nbsp;(Dexterous Hand Neural Dynamics Model) is a <b>groundbreaking neural dynamics model</b> that empowers robots to perform <b>high-precision in-hand rotations</b> of objects with varying sizes and geometries. From <b>tiny, delicate parts</b> to <b>large, irregular objects</b>, <b>DexNDM</b> adapts to dynamic forces and <b>varying wrist orientations</b>, solving a longstanding challenge in <b>dexterous manipulation</b>. This innovation pushes the boundaries of what robots can achieve in <b>fine motor skills</b>, offering <b>unparalleled flexibility</b> and <b>adaptability</b>.</p>  <p><b>Key highlights of DexNDM</b>:</p>  <ul type="disc">   <li><b>Unprecedented Generalization Across Object Types</b>: Unlike traditional models limited to specific object sizes or shapes, <b>DexNDM</b> can handle a wide range of objects, from <b>microelectronics</b> to <b>larger, more complex geometries</b> (such as books or elongated tools), breaking past limitations in robotic manipulation.</li>   <li><b>Multi-Axis, Multi-Posture Rotation</b>: <b>DexNDM</b> excels at rotating objects along <b>multiple axes</b>, regardless of the hand's orientation. Whether the hand is facing <b>up</b>, <b>down</b>, or sideways, the model ensures <b>stable and precise rotations</b>. This capability allows robots to perform complex tasks, such as adjusting the grip and rotating objects for precise <b>insertion</b> or <b>assembly</b>, even in hard-to-reach positions.</li>   <li><b>Highly Dexterous Teleoperation System</b>: Galbot has integrated <b>DexNDM</b> into a <b>teleoperation system</b>, enabling robots to perform <b>complex tasks</b> such as <b>screw tightening</b>, <b>hammering nails</b>, and <b>long-range assembly</b>. This system allows human operators to issue high-level commands while the robot autonomously handles <b>precise finger movements</b>, ensuring <b>unmatched flexibility</b> and <b>robustness</b>.</li>   <li><b>Expert-to-Generalist Learning Paradigm</b>: <b>DexNDM</b> uses a unique <b>expert-to-generalist learning approach</b>. The model first trains specialized policies for handling specific tasks and object types and then distills these policies into a unified, <b>adaptable strategy</b>. This approach allows <b>DexNDM</b> to seamlessly perform tasks across various robot types and environments with minimal fine-tuning.</li>  </ul>  <p><b>DexNDM</b>&nbsp;represents a significant milestone in the journey from robotic dexterity to <b>real-world productivity</b>, enabling robots to perform tasks like <b>tool use</b>, <b>assembly</b>, and <b>fine-grained manipulation</b> in industries like <b>manufacturing</b>, <b>healthcare</b>, and <b>logistics</b>.</p>  <p><b>NavFoM: A Leap in Autonomous Navigation</b></p>  <p><b>NavFoM</b>&nbsp;(Navigation Foundation Model) is the world's first <b>cross-embodiment, cross-task navigation foundation model</b> that enables robots to autonomously navigate across a wide range of environments and tasks. Developed by <b>Galbot</b> and its research collaborators, <b>NavFoM</b> consolidates fragmented navigation technologies into a unified system capable of seamlessly integrating <b>perception</b>, <b>understanding</b>, <b>decision-making</b>, and <b>action</b>.</p>  <p><b>Key highlights of NavFoM</b>:</p>  <ul type="disc">   <li><b>All Environments</b>: <b>NavFoM</b> supports both <b>indoor</b> and <b>outdoor navigation</b>, enabling <b>zero-shot</b> operation in <b>unseen environments</b> without requiring pre-mapping or additional training data. Robots powered by <b>NavFoM</b> can navigate new environments without prior knowledge, significantly reducing <b>deployment time</b> and <b>effort</b>. Whether navigating a <b>cluttered warehouse</b>, <b>crowded public space</b>, or <b>dynamic outdoor terrain</b>, <b>NavFoM</b> can seamlessly adapt to each scenario.</li>   <li><b>Cross-Task</b>: <b>NavFoM</b> can perform a wide range of navigation tasks, from <b>target-following</b> (e.g., following a person) to <b>autonomous navigation</b> (e.g., navigating through a busy urban area). It also works with <b>natural language commands</b>, allowing human operators to issue simple instructions, such as &quot;<b>follow that person</b>&quot; or &quot;<b>navigate to the red car</b>.&quot;</li>   <li><b>Cross-Embodiment</b>: One of <b>NavFoM's</b> standout features is its ability to adapt to various robot types, including <b>quadrupeds</b>, <b>wheeled humanoids</b>, <b>drones</b>, and even <b>vehicles</b>. This cross-embodiment adaptability ensures that <b>NavFoM</b> can be applied across different robot forms, enhancing <b>scalability</b> and <b>development efficiency</b>. Whether you're navigating an indoor robot, a drone, or an autonomous vehicle, <b>NavFoM</b> delivers consistent performance.</li>   <li><b>Unified Navigation Paradigm</b>: <b>NavFoM</b> introduces a novel method for processing <b>video streams</b> and <b>text commands</b> to generate <b>precise action trajectories</b>, unifying traditionally fragmented navigation models. This allows robots to <b>see</b>, <b>understand</b>, and <b>act</b> in real time across tasks and environments. For instance, a <b>quadruped robot's</b> ability to avoid crowds in a shopping mall can inform a <b>drone's</b> ability to navigate obstacles in the air, increasing <b>efficiency</b> and <b>accuracy</b> in unfamiliar environments.</li>  </ul>  <p>With <b>NavFoM</b>, <b>Galbot</b> has redefined the logic of navigation. Robots powered by <b>NavFoM</b> can <b>perceive</b> their surroundings, <b>plan paths</b>, and make real-time decisions in <b>completely unknown environments</b> with minimal human intervention.</p>  <p><b>Real-World Applications: Unlocking New Possibilities with&nbsp;DexNDM and NavFoM</b></p>  <p>Both <b>DexNDM</b> and <b>NavFoM</b> represent dual breakthroughs in <b>embodied AI</b>, pushing the boundaries of robotic capabilities in <b>manipulation</b> and <b>navigation</b>. <b>DexNDM</b> enhances robots' <b>dexterous manipulation</b> abilities, enabling high-precision, multi-step tasks involving tools and fine manipulation, while <b>NavFoM</b> redefines how robots <b>understand</b> and <b>navigate</b> their environments with <b>autonomy</b> and <b>adaptability</b>.</p>  <p>Together, these models provide a comprehensive solution for real-world applications across industries, from <b>assembly</b> and <b>tool use</b> in <b>manufacturing</b> to <b>navigation</b> and <b>autonomous decision-making</b> in <b>dynamic environments</b>.</p>  <p><b>A Future Powered by Embodied Intelligence</b></p>  <p>Galbot's <b>NavFoM</b> and <b>DexNDM</b> bring us closer to a world where robots not only <b>understand</b> their environment but can perform <b>complex tasks</b> with <b>precision</b> and <b>autonomy</b>. These innovations are poised to redefine industries, including <b>manufacturing</b>, <b>logistics</b>, <b>healthcare</b>, <b>retail</b>, and beyond, enabling robots to operate in real-world environments with <b>unprecedented adaptability</b> and <b>efficiency</b>.</p>  <p><b>For more information on DexNDM</b>, visit the <b>project website</b>: <u><a href="https://www.meowuu7.github.io/DexNDM" target="_blank" rel="nofollow" style="color: #0000FF">www.meowuu7.github.io/DexNDM</a></u></p>  <p><b>For more information on NavFoM</b>, visit the <b>project website</b>: <u><a href="https://pku-epic.github.io/NavFoM-Web/" target="_blank" rel="nofollow" style="color: #0000FF">https://pku-epic.github.io/NavFoM-Web/</a></u></p>  <p><b>About&nbsp;Galbot</b></p>  <p><b>Founded in 2023</b>, <b>Beijing Galbot Co., Ltd.</b> is a global leader in <b>embodied AI</b> and <b>general-purpose robotics</b>. Galbot's flagship product, the <b>Galbot G1</b>, has been widely deployed, achieving over <b>one year</b> of proven, stable real-world operations across industries such as <b>manufacturing</b>, <b>logistics</b>, <b>healthcare</b>, and <b>retail</b>. The <b>Galbot G1</b> is currently operating <b>more than 20 fully autonomous Galbot Stores</b> across over 20 cities, with plans to expand to <b>100 stores</b> by the end of the year.</p>  <p>With <b>over <span class="xn-money">$400M</span> in funding</b>, <b>Galbot</b> is transforming industries by setting new benchmarks in <b>automation</b>, delivering solutions that drive <b>operational efficiency</b>, <b>scalability</b>, and <b>productivity</b> across global sectors.</p>  <div class="PRN_ImbeddedAssetReference" id="DivAssetPlaceHolder2" align="center">   <p title="Experience NavFoM?셲 revolutionary navigation: With real-time adaptability and zero-shot generalization, NavFoM enables robots to autonomously navigate both familiar and unfamiliar environments, transforming how they perceive, understand, and make decisions in dynamic settings."><iframe width="560" height="315" src="https://www.youtube.com/embed/UMdQfnL7A3E" frameborder="0">&amp;amp;amp;amp;amp;nbsp;</iframe><br /><span>Experience NavFoM?셲 revolutionary navigation: With real-time adaptability and zero-shot generalization, NavFoM enables robots to autonomously navigate both familiar and unfamiliar environments, transforming how they perceive, understand, and make decisions in dynamic settings.</span></p>  </div>