PeopleofSuperintelligence

Researcher

Haotian Zhang

Haotian Zhang is a researcher whose work links 3D human motion modeling with humanoid robotics. He earned a B.E. in Computer Science and Technology from Tsinghua in 2017 and a computer science Ph.D. from Stanford in 2023. TextureNet was designed for high-resolution features on 3D meshes, and Vid2Player turned tennis broadcasts into controllable video sprites. In 2023, he coauthored a simulated-tennis system; in 2025, he coauthored GENMO, which unified motion estimation and generation; in 2026, he coauthored GRAIL, which used video model priors to create humanoid loco-manipulation data. As of October 2026, he is a Senior Research Scientist at NVIDIA, where he studies 3D human motion perception, modeling and generation for humanoid robotics.

Timeline

  1. 2017

    A Tsinghua computer science degree

    He earns a B.E. in Computer Science and Technology from Tsinghua University in 2017, advised by Shi-Min Hu.

  2. 2019

    TextureNet for 3D meshes

    In 2019, he coauthors TextureNet, a neural network designed to extract high-resolution features from 3D surface meshes. It uses a four-way rotationally symmetric field to define surface convolutions and supports semantic segmentation of textured meshes.

  3. 2021

    Controllable tennis video sprites

    In 2021, he coauthors Vid2Player, which converts annotated tennis broadcasts into interactively controllable video sprites. Its behavioral models use players' court positions and shot choices to select clips for match points.

  4. 2023

    A Stanford computer science PhD

    He completes a Ph.D. in Computer Science at Stanford in 2023, advised by Kayvon Fatahalian.

  5. 2023

    Simulated tennis from broadcasts

    In 2023, he coauthors a system that learns physically simulated tennis skills from broadcast videos. Its low-level imitation policy and high-level motion planner produce different shots and spins, and the SIGGRAPH paper receives a Best Paper Honorable Mention.

  6. 2025

    GENMO unifies motion tasks

    In 2025, he coauthors GENMO, a generalist model that unifies human motion estimation and generation. It handles variable-length motion with mixed text, audio and video conditions.

  7. June 2026

    GRAIL for humanoid robots

    In June 2026, he coauthors GRAIL, a digital pipeline that composes 3D assets, simulator-ready scenes and video foundation model priors to generate humanoid interactions. The team retargets recovered motions to a Unitree G1 humanoid and trains visual policies with data from the pipeline.

  8. As of October 2026

    Senior Research Scientist at NVIDIA

    As of October 2026, he is a Senior Research Scientist at NVIDIA. His research focuses on 3D human motion perception, modeling and generation, using human-motion videos to address humanoid robotics.

  1. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI from 1961 to 2025, with a year-by-year chart.
  2. The United States employs an estimated 53% of people in our data with a known employment countryOur estimates of where AI careers are concentrated now and how countries’ shares of dated positions changed over time.
  3. 46% of living people in our data with known birth and employment countries work outside their birth countryOur estimate of cross-border careers in AI, from country of birth and the country of each current organization.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionOur estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.

Sources and credits

Photo: We have no published photo of Haotian Zhang yet. Send us a free-licensed photo.

  1. Haotian Zhang | DAIR @ NVIDIA Research research.nvidia.com. Checked 19 October 2026.
  2. TextureNet: Consistent Local Parametrizations for Learning from High-Resolution Signals on Meshes web.stanford.edu. Checked 19 October 2026.
  3. Vid2Player: Controllable Video Sprites that Behave and Appear like Professional Tennis Players cs.stanford.edu. Checked 19 October 2026.
  4. Learning Physically Simulated Tennis Players from Broadcast Videos research.nvidia.com. Checked 19 October 2026.
  5. GENMO: A GENeralist Model for Human MOtion arxiv.org. Checked 19 October 2026.
  6. GRAIL: Generating Humanoid Loco-Manipulation from 3D Assets and Video Priors arxiv.org. Checked 19 October 2026.