Wenlong Huang

I am a Ph.D. candidate in Computer Science at Stanford University, working on AI and robotics. I am advised by Fei-Fei Li as part of the Stanford Vision and Learning Lab (SVL). I'm also working closely with Leslie Kaelbling and Tomás Lozano-Pérez.

I have spent time at MIT CSAIL (2025), NVIDIA Robotics (2024), and Google DeepMind Robotics (2022). I received my undergraduate degree from UC Berkeley (2018 - 2021), advised by Deepak Pathak, Igor Mordatch, and Pieter Abbeel. I also worked with Zhuowen Tu at UC San Diego as a summer intern (2018).

Email (wenlongh [AT] stanford.edu) Google Scholar Twitter GitHub LinkedIn

Follow @wenlong_huang

Research Highlights

3D World Model for Robotics [Mar 2026]

YouTube Watch on YouTube

Task Representations with Foundation Models [May 2025]

Task Representations with Foundation Models talk preview image

YouTube Watch on YouTube

My research focuses on robot learning, with the aim of developing algorithms with scalable data sources for broad generalization (e.g., across tasks _1,2,3,4, environments _1,2,3,4, and embodiments _1,2,3). Towards this goal, I am currently interested in:

Spatial Intelligence: Modeling the interactive, spatial, and counterfactual world - and the behaviors within it - for robotic manipulation.
Foundation Models for Robotics: Leveraging rich priors from foundation models and Internet-scale data for broad generalization.

News

[Jun 2026] Invited Talk at CVPR 2026 Workshop on Multi-Agent Robotic Systems: Scaling with Compositional Intelligence.
[Apr 2026] PointWorld is accepted at CVPR 2026 as a Highlight.
[Mar 2026] Invited Talk at Ai2, UT Austin, Lumina Talks, and The AI Talks on "3D World Models for Robotics".
[Feb 2026] Guest Lecture for AI-Robotics at University of Washington.
[Jun 2025] Co-organizing Workshop on Structured World Models for Robotic Manipulation at RSS 2025.
[May 2025] Invited Talk at ICRA 2025 Workshop on Foundation Models and Neuro-Symbolic AI for Robotics and Stanford Robotics Seminar.
[May 2025] UAD is nominated as a Best Paper Finalist and a Best Paper Award on Robot Perception Finalist at ICRA 2025.
[Apr 2025] Grateful to be selected as a finalist for the Citadel GQS Fellowship (three in CS/EE).
[Feb 2025] Co-organizing Tutorial on Foundation Models Meet Embodied Agents at ICCV 2025, NAACL 2025, and AAAI 2025.
[Feb 2025] Invited Talk at Amazon FAR, HKU on "Generalization through Task Representations with Foundation Models".
[Dec 2024] Grateful to be selected as a finalist for the NVIDIA Graduate Fellowship.
[Dec 2024] Co-organizing Workshop on Foundation Models for Embodied Agents (FMEA) at CVPR 2025.
[Dec 2024] Co-organizing Workshop on Planning in the Era of LLMs (LM4Plan) at AAAI 2025.
[Nov 2024] ReKep was presented at CoRL 2024 and won the Best Paper Award at the 2024 CoRL LEAP Workshop.
[Oct 2023] Invited Talks at Stanford IPRL, Tsinghua, TRI on "Foundation Models for Open-World Generalization in Robotics".
[Aug 2023] VoxPoser is accepted at CoRL 2023 for an Oral Presentation.
[Jun 2023] Code as Policies received Outstanding Robot Learning Paper Award at ICRA 2023.
[Nov 2022] Code as Policies is covered by Google AI Blog and TechCrunch.
[Oct 2022] Invited Talks at Columbia, Google on "Foundation Models 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.
[Feb 2022] Invited Talks at Google, FAIR, Sea AI Lab, ByteDance AI Lab on our LLMs as Zero-Shot Planners project.
[Dec 2021] Invited Talk at Intel AI Lab on "Generalization across Objects and Morphologies in Robot Learning".

Research

	PointWorld: Scaling 3D World Models for In-The-Wild Robotic Manipulation Wenlong Huang, Yu-Wei Chao, Arsalan Mousavian, Ming-Yu Liu, Dieter Fox, Kaichun Mo, Li Fei-Fei (Equal Advising) IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026 CVPR Highlight* Project Page / Paper / Code / Video / Summary A large 3D world model, pre-trained on 500 hours of in-the-wild 3D interactions, that predicts environment dynamics from RGB-D capture(s) and robot actions with unified state-action representation as 3D point flows.
	Dream2Flow: Bridging Video Generation and Open-World Manipulation with 3D Object Flow Karthik Dharmarajan, Wenlong Huang, Jiajun Wu, Li Fei-Fei†, Ruohan Zhang† (Corresponding Author, †Equal Advising) International Conference on Robotics and Automation (ICRA), 2026 Project Page / Paper / Code / Summary 3D object flows from text-to-video models provide zero-shot guidance for open-world manipulation as tracking goals for downstream (model-based and model-free) policies.
	Learning Composable Skills by Discovering Spatial and Temporal Structure with Foundation Models Neil Nie, Wenlong Huang, Jiayuan Mao, Li Fei-Fei, Weiyu Liu, Jiajun Wu International Conference on Robotics and Automation (ICRA), 2026 Project Page / Paper Extract skill segments and 3D entities from unstructured demonstrations to learn composable skills that generalize to longer-horizon test settings and novel geometric constraints.
	ENACT: Evaluating Embodied Cognition with World Modeling of Egocentric Interaction Qineng Wang, Wenlong Huang, Yu Zhou, Hang Yin, Tianwei Bao, Jianwen Lyu, Weiyu Liu, Ruohan Zhang†, Jiajun Wu†, Li Fei-Fei†, Manling Li† (Equal Contribution, †Equal Advising) International Conference on Learning Representations (ICLR), 2026 Oral Presentation* at the ICLR 2026 Workshop on World Models Project Page / Paper / Code / Dataset / Summary A benchmark that evaluates embodied cognition of VLMs on long-horizon, spatial, and physical reasoning through egocentric world modeling from long-horizon mobile manipulation.
	ReKep: Spatio-Temporal Reasoning of Relational Keypoint Constraints for Robotic Manipulation Wenlong Huang, Chen Wang, Yunzhu Li, Ruohan Zhang, Li Fei-Fei (Equal Contribution) The Conference on Robot Learning (CoRL), 2024 Best Paper Award* at the 2024 CoRL LEAP Workshop Project Page / Paper / Code / Video / Summary Large vision models and vision-language models can generate keypoint-based constraints, which can be optimized to achieve multi-stage, in-the-wild, bimanual, and reactive behaviors, without task-specific training or environment models.
	UAD: Unsupervised Affordance Distillation for Generalization in Robotic Manipulation Yihe Tang, Wenlong Huang, Yingke Wang, Chengshu Li, Roy Yuan, Ruohan Zhang, Jiajun Wu, Li Fei-Fei (Corresponding Author) International Conference on Robotics and Automation (ICRA), 2025 Best Paper Award Finalist Best Paper Award on Robot Perception Finalist Project Page / Paper / Code / Summary Fine-grained task-conditioned visual affordance can be distilled from off-the-shelf foundation models, enabling diverse generalization properties in downstream policy learning.
	A Real-to-Sim-to-Real Approach to Robotic Manipulation with VLM-Generated Iterative Keypoint Rewards Shivansh Patel, Xinchen Yin, Wenlong Huang, Shubham Garg, Hooshang Nayyeri, Li Fei-Fei, Svetlana Lazebnik, Yunzhu Li (Equal Contribution) International Conference on Robotics and Automation (ICRA)*, 2025 Project Page / Paper / Code / Video / Summary Vision-Language Models can serve as human proxies specify diverse task objectives by writing keypoint-based reward functions, where they can be autonomously and iteratively refined based on environment feedback.
	VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models Wenlong Huang, Chen Wang, Ruohan Zhang, Yunzhu Li, Jiajun Wu, Li Fei-Fei The Conference on Robot Learning (CoRL), 2023 Oral Presentation Project Page / Paper / Code / Video / Summary Large language models and visual-language models can be used to directly label affordances and constraints in the 3D perceptual space. Combined with motion planning, this enables robots to perform diverse everyday manipulation tasks in a zero-shot manner.
	PaLM-E: An Embodied Multimodal Language Model Danny Driess, Fei Xia, Mehdi S. M. Sajjadi, Corey Lynch, Aakanksha Chowdhery, Brian Ichter, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, Wenlong Huang, Yevgen Chebotar, Pierre Sermanet, Daniel Duckworth, Sergey Levine, Vincent Vanhoucke, Karol Hausman, Marc Toussaint, Klaus Greff, Andy Zeng, Igor Mordatch, Pete Florence International Conference on Machine Learning (ICML), 2023. Project Page / Paper / Google AI Blog / Summary Language models can digest real-world sensor modalities (e.g., images) to be embodied in the physical world. The largest 562B model is a generalist agent across language, vision, and task planning.
	Grounded Decoding: Guiding Text Generation with Grounded Models for Robot Control Wenlong Huang, Fei Xia, Dhruv Shah, Danny Driess, Andy Zeng, Yao Lu, Pete Florence, Igor Mordatch, Sergey Levine, Karol Hausman, Brian Ichter Conference on Neural Information Processing Systems (NeurIPS), 2023 Project Page / Paper / Video / Summary Large language models can be grounded in embodied environments by using continuous probabilities to guide their token decoding, where the guidance is provided by a set of grounded models, such as affordance, safety, and preference functions.
	Code as Policies: Language Model Programs for Embodied Control Jacky Liang, Wenlong Huang, Fei Xia, Peng Xu, Karol Hausman, Brian Ichter, Pete Florence, Andy Zeng International Conference on Robotics and Automation (ICRA), 2023 Outstanding Robot Learning Paper Award Project Page / Paper / Code / Video / Google AI Blog / TechCrunch / Summary Using hierarchical code generation, language models can write hierarchical robot policy code given abstract natural language instructions.
	Inner Monologue: Embodied Reasoning through Planning with Language Models Wenlong Huang, Fei Xia, Ted Xiao, Harris Chan, Jacky Liang, Pete Florence, Andy Zeng, Jonathan Tompson, Igor Mordatch, Yevgen Chebotar, Pierre Sermanet, Noah Brown, Tomas Jackson, Linda Luu, Sergey Levine, Karol Hausman, Brian Ichter (Equal Contribution) The Conference on Robot Learning (CoRL), 2022 Project Page / Paper / Video / Two Minute Papers / Summary Provided with textual embodied feedback, language models can articulate a grounded "thought process" for challenging long-horizon tasks, even under disturbances.
	Language Models as Zero-Shot Planners: Extracting Actionable Knowledge for Embodied Agents Wenlong Huang, Pieter Abbeel, Deepak Pathak, Igor Mordatch (Equal Advising) International Conference on Machine Learning (ICML)*, 2022. Project Page / Paper / Code / Video / Summary Large language models (e.g. GPT-3, Codex) contain rich actionable knowledge that can be used to perform task planning for embodied agents, even without additional training.
	Generalization in Dexterous Manipulation via Geometry-Aware Multi-Task Learning Wenlong Huang, Igor Mordatch, Pieter Abbeel, Deepak Pathak arXiv, 2021 Project Page / Paper / Code / Summary With appropriate object representation, a multi-task RL policy can control an anthropomorphic hand to manipulate 100+ diverse objects and achieve SOTA performance on unseen ones.
	One Policy to Control Them All: Shared Modular Policies for Agent-Agnostic Control Wenlong Huang, Igor Mordatch, Deepak Pathak International Conference on Machine Learning (ICML), 2020. Oral Presentation Project Page / Paper / Code / Video / Oral Talk / Summary Expressing robots as collections of modular components that share a control policy can lead to zero-shot generalization across diverse unseen robot morphologies.
	3D Volumetric Modeling with Introspective Neural Networks Wenlong Huang, Brian Lai, Weijian Xu, Zhuowen Tu (Equal Contributions) Association for the Advancement of Artificial Intelligence (AAAI)*, 2019. Paper Built upon Generative via Discriminative Learning and Introspective Learning frameworks, a single neural network can simultaneously perform classification and generation of 3D volumetric shapes.