<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Zhizheng Liu's homepage</title><link>https://bosmallear.github.io/</link><description>Recent content on Zhizheng Liu's homepage</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Mon, 16 Nov 2026 16:38:42 +0000</lastBuildDate><atom:link href="https://bosmallear.github.io/index.xml" rel="self" type="application/rss+xml"/><item><title>Readme</title><link>https://bosmallear.github.io/readme/</link><pubDate>Wed, 22 Feb 2023 20:03:14 +0100</pubDate><guid>https://bosmallear.github.io/readme/</guid><description>&lt;p&gt;&lt;strong&gt;Hi, my name is Zhizheng Liu (刘知正)&lt;/strong&gt;. I am currently a fourth-year Ph.D. candidate in Computer Science at &lt;a href="https://vail-ucla.github.io/"&gt;Vision and Autonomy Intelligence Lab (VAIL)&lt;/a&gt; at UCLA. Before that, I earned my bachelor’s degree from the SJTU-UM Joint Institute (now SJTU Global College) and my master’s degree from ETH Zurich.&lt;/p&gt;&#10;&lt;p&gt;I work on a fundamental problem in embodied AI: &lt;em&gt;&lt;strong&gt;How to enable general intelligent agents to understand and interact with humans in the physical world?&lt;/strong&gt;&lt;/em&gt;. To address this challenging problem, my PhD research focuses on closing the real-to-sim gap for human-centric robotics with &lt;strong&gt;data-driven simulation&lt;/strong&gt;. To this end, my work involves &lt;a href="https://vail.cs.ucla.edu/JOSH/"&gt;monocular 4D human-scene reconstruction&lt;/a&gt;, using the reconstructed data for &lt;a href="https://genforce.github.io/PedGen/"&gt;generative modeling of pedestrian motion&lt;/a&gt;, and transferring these digital twins into simulation environments for scalable robot &lt;a href="https://metadriverse.github.io/vid2sim/"&gt;training&lt;/a&gt; and &lt;a href="https://vail.cs.ucla.edu/SidewalkBench/"&gt;evaluation&lt;/a&gt;.&lt;/p&gt;</description></item></channel></rss>