Researcher

Dhruv Shah

Dhruv Shah is a robotics researcher whose work combines machine learning and foundation models for robot navigation and control. He holds a Ph.D. in electrical engineering and computer science from UC Berkeley, where Sergey Levine advised him, and a B.Tech. in electrical engineering from IIT Bombay. LM-Nav and ViNT brought pre-trained models to robotic navigation, while his 2024 Berkeley dissertation proposed foundation models for open-world robots. His goal-masked diffusion paper won an IEEE ICRA Best Conference Paper Award in 2024, and he coauthored the Gemini Robotics report for direct robot control. He previously co-led Gemini Robotics at Google DeepMind; as of October 2026, he is an assistant professor of robotics at Princeton and a research scientist at Google DeepMind.

Timeline

  1. 2022

    An RSS systems-paper finalist

    In 2022, his research is a finalist for the Best Systems Paper Award at Robotics: Science and Systems.

  2. 2023

    LM-Nav and language instructions

    At CoRL in 2023, he coauthors LM-Nav, which uses pre-trained navigation, image-language and language models to give users a high-level interface to robot navigation without fine-tuning or language-annotated robot data.

  3. 2023

    ViNT visual navigation

    He coauthors ViNT in 2023, a Transformer-based foundation model for vision-based robotic navigation that adapts to downstream tasks.

  4. 2024

    A dissertation on open-world robots

    He publishes his UC Berkeley dissertation, The Foundation Model Path to Open-World Robots, in 2024. It proposes general robot foundation models trained on data from different environments and embodiments, using egocentric vision.

  5. 2024

    A goal-masked diffusion award

    In 2024, he coauthors Goal Masked Diffusion Policies for Unified Navigation and Exploration, which wins IEEE ICRA's Best Conference Paper Award.

  6. 2024

    Microsoft Future Leader

    He is a Microsoft Future Leader in Robotics & AI in 2024.

  7. March 2025

    Gemini Robotics for direct control

    In March 2025, he coauthors the Gemini Robotics report, which introduces a family of Gemini 2.0-based models and a vision-language-action model for direct robot control.

  8. July 2026

    Whole-body control in Gemini Robotics 2

    In July 2026, he appears among the Gemini Robotics 2 team members credited for a model that controls humanoid robots with whole-body movement.

  9. As of October 2026

    Princeton robotics faculty

    His education includes a Ph.D. in EECS at UC Berkeley, where Sergey Levine advised him, and a B.Tech. in electrical engineering at IIT Bombay. He has spent time at Google Brain and Meta FAIR and previously co-led Gemini Robotics at Google DeepMind, scaling robotic intelligence for humanoids. As of October 2026, he is an assistant professor of robotics at Princeton, where he leads the PRISM group, and a research scientist at Google DeepMind.

Career

  1. NowGoogle DeepMindResearch Scientist
  2. NowPrinceton UniversityAssistant Professor of Robotics

Education

  1. 2024University of California, BerkeleyPh.D.
  2. IIT BombayB.Tech.

Sources and credits

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

  1. Microsoft Future Leaders in Robotics and AI Seminar Series: Dhruv Shah robotics.umd.edu. Checked 26 October 2026.
  2. LM-Nav: Robotic Navigation with Large Pre-Trained Models of Language, Vision, and Action proceedings.mlr.press. Checked 26 October 2026.
  3. ViNT: A Foundation Model for Visual Navigation proceedings.mlr.press. Checked 26 October 2026.
  4. The Foundation Model Path to Open-World Robots www2.eecs.berkeley.edu. Checked 26 October 2026.
  5. Awards and Finalists - 2024 IEEE International Conference on Robotics and Automation (ICRA 2024) 2024.ieee-icra.org. Checked 26 October 2026.
  6. Gemini Robotics: Bringing AI into the Physical World arxiv.org. Checked 26 October 2026.
  7. Gemini Robotics 2 brings whole body intelligence to robots deepmind.google. Checked 26 October 2026.
  8. Dhruv Shah dhruvshah.me. Checked 26 October 2026.