I am the Co-founder and Chief Scientist at SimpleAI, where I lead research on world models and continual learning for general embodied intelligence. We are building robots that learn continuously through interaction, understand the physical world, and generalize across tasks, environments, and embodiments, with an initial focus on next-generation home robots.
Previously, I was a Shuimu Scholar and Postdoctoral Researcher at the School of Vehicle and Mobility, Tsinghua University, where I also received my Ph.D. under the supervision of Prof. Diange Yang. My earlier research treated autonomous driving as a safety-critical embodied AI problem, spanning data loops, model and policy training, closed-loop evaluation, and online improvement.
Our work on continual improvement for self-driving cars, published in Nature Machine Intelligence, was applied in the autonomous-driving demonstration for the 2022 Beijing Winter Olympics, which operated with zero accidents. Related systems have been deployed at scale at Didi Autonomous Driving and Toyota, with research collaborations involving Baidu Apollo and XPeng.
Additional highlights include publications at ICML, IROS, and IEEE T-ITS; Didi’s Gaia Lighthouse Outstanding Project Award and Most Popular Project Award; and winning the Mcity AV Challenge, where our autonomous vehicle completed the competition without a collision.
Follow our latest work at Simple World Lab. We are hiring—please reach out if you are interested in joining us. Email: zhouwt801 [at] gmail [dot] com.
🔥 News
- 2026.07: We are excited to introduce HiFi-UMI: Learning Deployable Manipulation Policies from High-Fidelity UMI Data Alone. HiFi-UMI enables policies trained solely on portable, robot-free demonstrations to deploy directly on real robots, and we are releasing a 2,000-hour high-fidelity dataset to support scalable robot learning.
- 2026.07: I am co-organizing the Safe World Models for Trustworthy Embodied AI workshop at ECCV 2026, bringing together researchers working on reliable and safe world models for embodied agents.
- 2026.07: We are excited to share two new embodied-AI studies: Diagnosing Semantic Handoff Failures in Agent-Orchestrated Vision-Language-Action Skill Composition and SoftVTBench: A Safety-Aware Visuo-Tactile Benchmark for Physically Constrained Robotic Manipulation of Deformable Objects.
- 2026.06: I am co-organizing the Physical World Models for Scaling Embodied AI workshop at IROS 2026, focusing on how physical world models can help scale embodied intelligence.
- 2026.06: Our paper Dynamics Are Learned, Not Told: Semi-Supervised Discovery of Latent Dynamics Geometries for Zero-Shot Policy Adaptation will appear at ICML 2026.
- 2026.05: Happy to share our new paper GaussianDream: A Feed-Forward 3D Gaussian World Model for Robotic Manipulation, which explores efficient 3D world modeling for robot learning.
- 2026.03: Our paper CounterScene: Counterfactual Causal Reasoning in Generative World Models for Safety-Critical Closed-Loop Evaluation is now online.
- 2025.09: Our work on long-tail autonomous driving, developed at Tsinghua and deployed on Didi’s RoboTaxi, received Didi’s top honor—the Gaia Lighthouse Outstanding Project Award—as well as the employee-voted Most Popular Project Award. News
- 2025.07: We organized the Next Generation of Self-Driving Vehicles forum at Tsinghua University, featuring invited speakers from academia and industry. Talk videos
- 2025.06: Our paper DRARL: Disengagement-Reason-Augmented Reinforcement Learning for Efficient Improvement of Autonomous Driving Policy was accepted by IROS 2025.
- 2025.03: Our paper Dynamically Local-Enhancement Planner for Large-Scale Autonomous Driving is now online.
- 2024.10: We organized the Multi-Agent Autonomous Systems Workshop at ECCV 2024.
- 2024.09: We organized the invited session Driving the Edge: Addressing Corner Cases in Self-Driving Vehicles at ITSC 2024.
- 2024.09: We won the Mcity AV Challenge, and our autonomous vehicle completed the entire competition without a collision.
- 2024.03: We released SPIDER, an open-source toolkit for building reusable data-driven and rule-based self-driving planners.
- 2023.03: Our work Continuous Improvement of Self-Driving Cars Using Dynamic Confidence-Aware Reinforcement Learning was published in Nature Machine Intelligence.
📝 Publications
Selected work is organized by research theme; see my Google Scholar profile for the complete, up-to-date list.
Embodied AI & World Models
Diagnosing Semantic Handoff Failures in Agent-Orchestrated Vision-Language-Action Skill Composition
Ke Rui, Yushen Zuo, Jiawei Wang, Haoran Jia, Jinming Ma, Weitao Zhou, Minglei Li
SoftVTBench: A Safety-Aware Visuo-Tactile Benchmark for Physically Constrained Robotic Manipulation of Deformable Objects
Bowen Jing, Mingxin Wang, Ruiyang Hao, Chenchen Ge, Hanwen Shen, Junjie He, Yang Cui, Yiming Hou, Weitao Zhou, et al.
Dynamics Are Learned, Not Told: Semi-Supervised Discovery of Latent Dynamics Geometries for Zero-Shot Policy Adaptation
Zhiming Xu, Weitao Zhou, Xianghui Pan, Nanshan Deng, Chengju Liu, Qijun Chen, Chenpeng Yao
GaussianDream: A Feed-Forward 3D Gaussian World Model for Robotic Manipulation
Zijian Zhang, Yuqing Jiang, Qian Cheng, Xiaofan Li, Si Liu, Ding Zhao, Ping Luo, Weitao Zhou, Haibao Yu
CounterScene: Counterfactual Causal Reasoning in Generative World Models for Safety-Critical Closed-Loop Evaluation
Bowen Jing, Ruiyang Hao, Weitao Zhou, Haibao Yu
Continual Learning & Reinforcement Learning
Dual-Flow Reinforcement Learning with State-Aware Exploration
Qijun Li, Zheng Fu, Qi Song, Yifei He, Weitao Zhou, Kun Jiang, Diange Yang
DRARL: Disengagement-Reason-Augmented Reinforcement Learning for Efficient Improvement of Autonomous Driving Policy
Weitao Zhou, Bo Zhang, Zhong Cao, Xiang Li, Qian Cheng, Chunyang Liu, Yaqin Zhang, Diange Yang
Continuous Improvement of Self-Driving Cars Using Dynamic Confidence-Aware Reinforcement Learning
Zhong Cao, Kun Jiang, Weitao Zhou, Shaobing Xu, Huei Peng, Diange Yang · Project
Identify, Estimate and Bound the Uncertainty of Reinforcement Learning for Autonomous Driving
Weitao Zhou, Zhong Cao, Nanshan Deng, Kun Jiang, Diange Yang
Autonomous Driving
Dynamically Local-Enhancement Planner for Large-Scale Autonomous Driving
Nanshan Deng, Weitao Zhou, Bo Zhang, Junze Wen, Kun Jiang, Zhong Cao, Diange Yang
SPIDER: Self-Driving Planners and Intelligent Decision-Making Engines with Reusability
Zelin Qian, Kun Jiang, Zhong Cao, Kai Qian, Yunkang Xu, Weitao Zhou, Diange Yang
Dynamically Conservative Self-Driving Planner for Long-Tail Cases
Weitao Zhou, Zhong Cao, Nanshan Deng, Xiaoyu Liu, Kun Jiang, Diange Yang
Long-Tail Prediction Uncertainty Aware Trajectory Planning for Self-Driving Vehicles
Weitao Zhou, Zhong Cao, Yunkang Xu, Nanshan Deng, Xiaoyu Liu, Kun Jiang, Diange Yang
Integrating Deep Reinforcement Learning with Optimal Trajectory Planner for Automated Driving
Weitao Zhou, Kun Jiang, Zhong Cao, Nanshan Deng, Diange Yang
💬 Academic Service
- Reviewer: Nature Machine Intelligence, Nature Computational Science, IEEE T-ITS, IEEE T-IV, ICRA, ITSC, IROS, ECCV, NeurIPS, ICML, CVPR.
- Workshop Organizer:
- ECCV 2026 — Safe World Models for Trustworthy Embodied AI
- IROS 2026 — Physical World Models for Scaling Embodied AI
- ECCV 2024 — Multi-Agent Autonomous Systems Workshop
- ITSC 2024 · Invited Session — Driving the Edge: Addressing Corner Cases in Self-Driving Vehicles
- Forum Organizer:
- Tsinghua University · 2025 — Next Generation of Self-Driving Vehicles
- SAECCE 2023 — Autonomous-Driving Map Updates and Safety Compliance
💼 Experience
Co-founder & Chief Scientist · SimpleAI
Leading research on world models and continual learning for general embodied intelligence and next-generation home robots.
Postdoctoral Researcher · Tsinghua University
Shuimu Scholar at the School of Vehicle and Mobility, developing continual-learning and trustworthy autonomous-driving systems.
Research Intern · Didi Autonomous Driving
Conducted research on reinforcement learning and long-tail autonomous-driving policy improvement.
Research Intern · Baidu Apollo
Worked on motion planning and decision-making for autonomous driving.
Research Intern · Idriverplus
Worked on planning and control for autonomous vehicles.