<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: Razied</title><link>https://news.ycombinator.com/user?id=Razied</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Tue, 14 Apr 2026 20:18:49 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=Razied" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by Razied in "Show HN: Train and deploy your own open-source humanoid in Python"]]></title><description><![CDATA[
<p>Super cool library guys, I think this is by far the quickest way and most painless way to train a humanoid policy. I tried messing around with mujoco and isaaclab stuff a while ago, and it was truly horrible.<p>A bit more of a hardware question: how modular are the robots? I'm looking at the feet and hands in particular and thinking that there are a bunch of applications where purpose-build parts would work much better than the current ones you have there. I understand the arguments for a humanoid form-factor, but I think flat feet are making it much harder than it should be to get a robust locomotion policy, and training hand-dexterity is also unbelievably hard. Seems like the path of least resistance to usefulness is to have an array of different hand attachments.</p>
]]></description><pubDate>Sun, 18 May 2025 17:41:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=44023031</link><dc:creator>Razied</dc:creator><comments>https://news.ycombinator.com/item?id=44023031</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44023031</guid></item></channel></rss>