Lester Li

Contact: sizheli [at] mit [dot] edu

I am a PhD student in Computer Science at MIT CSAIL, advised by Vincent Sitzmann and Josh Tenenbaum . My research is supported by the MIT Presidential Fellowship.

Research Objectives. I study how robots can learn to model their bodies and the physical world, predict the consequences of their actions, and use these predictions for planning and control. My work connects representation learning, generative modeling, and physical reasoning, with the goal of enabling robots to generalize across tasks, environments, and embodiments.

News

May 27, 2026 We released VERA, a video-model-based generalist robot policy. Read the paper and the announcement.
Jul 07, 2025 Thank you Forbes, MIT News, and DeepTech深科技 for covering our recent work!
Mar 23, 2025 Very excited to share my blog post on robot modeling and representation! A tutorial of Jacobian Fields with code.
Jul 15, 2024 Check out our new preprint on learning representation of robotic embodiment!

Blog

Student Mentorship

Selected Publications

All publications
  1. arXiv 2026

    VERA: Turning Video Models into Generalist Robot Policies

    Sizhe Lester Li*, Evan Kim*, Xingjian Bai*, Tong Zhao, Tao Pang, Max Simchowitz, and Vincent Sitzmann
    * Equal contribution.
  2. Nature 2025

    Controlling diverse robots by inferring Jacobian fields with deep networks

    Sizhe Lester Li, Annan Zhang, Boyuan Chen, Hanna Matusik, Chao Liu, Daniela Rus, and Vincent Sitzmann
  3. CVPR 2024 Best Paper, Runners-Up

    pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction

    David Charatan, Sizhe Lester Li, Andrea Tagliasacchi, and Vincent Sitzmann
  4. ICLR 2023

    DexDeform: Dexterous Deformable Object Manipulation with Human Demonstrations and Differentiable Physics

    Sizhe Lester Li*, Zhiao Huang*, Tao Chen, Tao Du, Hao Su, Joshua B. Tenenbaum, and Chuang Gan
  5. ICLR 2022 Spotlight Presentation

    Contact Points Discovery for Soft-Body Manipulations with Differentiable Physics

    Sizhe Lester Li*, Zhiao Huang*, Tao Du, Hao Su, Joshua Tenenbaum, and Chuang Gan