Wenlong Huang

I am a first-year Ph.D. student in Computer Science at Stanford University.

I recently received my B.A. in Computer Science from UC Berkeley, where I was fortunate to be advised by Deepak Pathak, Igor Mordatch, and Pieter Abbeel. After graduation, I interned in the Robotics team at Google Brain. Before my time at Berkeley, I was also fortunate to work with Zhuowen Tu at UC San Diego.

Email  /  Google Scholar  /  Twitter  /  GitHub

• [Nov 2022] Code as Policies is covered by Google AI Blog and TechCrunch.
• [Oct 2022] Invited Talks @ CAIR, Google on language as generalization interface for robotics.
• [Oct 2022] Inner Monologue is covered by Two Minute Papers.
• [Sep 2022] Started at Stanford as a PhD student, generously supported by Stanford SoE Fellowship.
• [Apr 2022] Started my internship at Google Brain as a student researcher, working with the amazing Robotics team.
• [Feb 2022] Interviewed by Yannic Kilcher for our language planner project. Check it out here!
• [Feb 2022] Invited Talks @ Google, FAIR, Sea AI Lab, ByteDance AI Lab on our language planner project.
• [Dec 2021] Invited Talk @ Intel AI Lab on generalization across objects and morphologies in robot learning.

I'm broadly interested in robot learning. The goal of my research is to build agents that can make intelligent decisions in embodied environments and have generalizable motor skills in challenging scenarios. Recently, I am interested in leveraging large pre-trained models to improve generalization of robot capabilities.

Language models can digest real-world sensor modalities (e.g., images) to be embodied in the physical world. The largest model with 562B parameters is a generalist agent across language, vision, and robot planning.

We formulate a token decoding procedure applying large language model to robotics settings. Tokens are selected based on likelihood under the language model and a set of grounded models, such as affordance, safety, and preference functions.

Using hierarchical code generation, large language models can write robot policy code that exhibits spatial-geometric reasoning when given abstract natural language instructions, without any additional training.

Using various sources to provide textual embodied feedback, frozen large language models can articulate a grounded "thought process" for robots, solving many challenging long-horizon robotics tasks, even under adversarial perturbation.

Large language models (e.g. GPT-3, Codex) contain rich actionable knowledge that can be used to plan actions for embodied agents, even without additional training.

With appropriate object representation, a multi-task reinforcement learning policy can control an anthropomorphic hand to manipulate 100+ diverse objects and achieve SOTA performance on unseen ones.

Expressing robots as collections of modular components that share a control policy can lead to zero-shot generalization across diverse unseen robot morphologies.

Built upon prior Generative via Discriminative Learning and Introspective Learning frameworks, a single neural network can simultaneously perform classification and generation of 3D volumetric shapes.

Reviewer for ICML, IROS, NeurIPS.

Template from Jon Barron's website