<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/95c7d0a9ccb74c4a8131b5a015a7dc42&quot; frameborder=&quot;0&quot; width=&quot;1286&quot; height=&quot;964&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>964</height><width>1286</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>964</thumbnail_height><thumbnail_width>1286</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/95c7d0a9ccb74c4a8131b5a015a7dc42-00001.gif</thumbnail_url><duration>343.9000000000005</duration><title>The Future of Robot Control</title><description>In this video, I demonstrate the incredible potential of robot learning. I showcase how we can automate physical tasks using a specific robot arm and a set-up. By training a generalist control policy with diverse robot data, we aim to input tasks in natural language and have the control policy effectively control the robot. I discuss the challenges and breakthroughs in this field, including the development of hardware setups, the release of large-scale robotic datasets, and the use of pre-trained policies. Join me to explore the exciting possibilities of robot learning!</description></oembed>