Andy Zeng

About

Andy Zeng is a leading robotics and AI researcher whose work sits at the intersection of robot manipulation, computer vision, and large language models (LLMs). His research has fundamentally advanced how robots perceive, reason about, and interact with the physical world—spanning grasping, throwing, navigation, and language-conditioned control. Zeng's early contributions tackled core manipulation challenges, including 6D pose estimation for warehouse automation (487 citations) and multi-affordance grasping of novel objects in cluttered scenes (461 citations), helping lay the groundwork for practical robotic pick-and-place systems. His creative work on TossingBot (284 citations) demonstrated that robots could learn to throw arbitrary objects accurately by combining residual physics with deep learning, while ClearGrasp (258 citations) addressed the notoriously difficult problem of manipulating transparent objects. More recently, Zeng has been a central figure in bridging LLMs with embodied robotics. Landmark papers such as Code as Policies (561 citations), PaLM-E (350 citations), Inner Monologue (206 citations), and VLMaps (301 citations) collectively show how language models can ground natural language commands into executable robot behavior. With over 3,000 citations across his most recognized works, Zeng's research is shaping the future of intelligent, language-capable robotic systems.

Research Focus

Key Achievements

Top Papers

1. 1
   Code as Policies: Language Model Programs for Embodied Control

   561 citations · 2023
2. 2
   Multi-view self-supervised deep learning for 6D pose estimation in the Amazon Picking Challenge

   487 citations · 2017
3. 3
   Robotic Pick-and-Place of Novel Objects in Clutter with Multi-Affordance Grasping and Cross-Domain Image Matching

   461 citations · 2018
4. 4
   PaLM-E: An Embodied Multimodal Language Model

   350 citations · 2023
5. 5
   Visual Language Maps for Robot Navigation

   301 citations · 2023
6. 6
   TossingBot: Learning to Throw Arbitrary Objects With Residual Physics

   284 citations · 2020
7. 7
   Clear Grasp: 3D Shape Estimation of Transparent Objects for Manipulation

   258 citations · 2020
8. 8
   Inner Monologue: Embodied Reasoning through Planning with Language Models

   206 citations · 2022
9. 9
   Robotic pick-and-place of novel objects in clutter with multi-affordance grasping and cross-domain image matching

   198 citations · 2019
10. 10
    TidyBot: personalized robot assistance with large language models

    189 citations · 2023

Key Collaborators

Contact & Links