ZhenghaoPeng-2024.jpg

ZhenghaoPeng-2024.jpg
Zhenghao Peng

Hi there! I’m a final-year PhD student in Computer Science at UCLA, advised by Professor Bolei Zhou. Before UCLA, I earned my MPhil in the Multimedia Lab at the Chinese University of Hong Kong, and my Bachelor’s at Shanghai Jiao Tong University.

I am building agents that can reason about the world, align with human intent, and adapt in real time. Over the past 7 years, I’ve explored a wide spectrum of agent learning techniques: multi-agent RL, human-in-the-loop learning, and large-scale training.

Goal: Physically-grounded, aligned and versatile robots.

Method: Imitation learning pretraining, closed-loop finetuning, and human-in-the-loop test-time adaptation.

Please check out my research statement for more details.

I’m currently on the job market — if you’re working on the future of embodied AI, let’s chat!



Recent News

Jun 16, 2025 I’ll be interning at the Autonomous Vehicles Research Group at NVIDIA this summer.
Jun 06, 2025 I received the Dissertation Year Award! Thanks, UCLA!
Feb 26, 2025 Papers on building RL env from video via 3D GS and improving VLM via MetaVQA were accepted to CVPR 2025.
Jan 27, 2025 Paper on applying human-in-the-loop learning on real robots was accepted to ICRA 2025.
Sep 25, 2024 Papers on shared autonomy (AI assists human) and diffusion on driving were accepted to NeurIPS 2024.
Jul 03, 2024 Paper on RL finetuning behavior model was accepted to ECCV 2024.
Jun 14, 2024 I am honored to receive the Amazon Fellowship. Many thanks to Amazon!
Sep 21, 2023 Human-in-the-loop learning method PVP was accepted to NeurIPS 2023 as Spotlight! ScenarioNet was accepted to NeurIPS 2023 Dataset Track!

Selected Projects

  1. arXiv
    InfGen: Scenario Generation as Next-Token-Group Prediction
    Zhenghao Peng, Yuxin Liu, and Bolei Zhou
    In Preprint, 2025
  2. ICRA
    Data-Efficient Learning from Human Interventions for Mobile Robots
    Zhenghao Peng, Zhizheng Liu, and Bolei Zhou
    In International Conference on Robotics and Automation, 2025
  3. NeurIPS
    Shared Autonomy with IDA: Interventional Diffusion Assistance
    Brandon J. McMahan, Zhenghao Peng, Bolei Zhou, and Jonathan C. Kao
    Advances in Neural Information Processing Systems, 2024
  4. ECCV
    Improving Agent Behaviors with RL Fine-tuning for Autonomous Driving
    Zhenghao Peng, Wenjie Luo, Yiren Lu, Tianyi Shen, Cole Gulino, Ari Seff, and Justin Fu
    In European Conference on Computer Vision, 2024
  5. NeurIPS Spotlight
    Learning from Active Human Involvement through Proxy Value Propagation
    Zhenghao Peng, Wenjie Mo, Chenda Duan, Quanyi Li, and Bolei Zhou
    In Advances in Neural Information Processing Systems, 2023
  6. NeurIPS
    ScenarioNet: Open-Source Platform for Large-Scale Traffic Scenario Simulation and Modeling
    Quanyi Li*Zhenghao Peng*, Lan Feng*, Zhizheng Liu, Chenda Duan, Wenjie Mo, and Bolei Zhou
    In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2023
  7. CoRL
    CAT: Closed-loop Adversarial Training for Safe End-to-End Driving
    Linrui Zhang, Zhenghao Peng, Quanyi Li, and Bolei Zhou
    In 7th Annual Conference on Robot Learning, 2023
  8. ICLR Spotlight
    Guarded Policy Optimization with Imperfect Online Demonstrations
    Zhenghai Xue, Zhenghao Peng, Quanyi Li, Zhihan Liu, and Bolei Zhou
    In International Conference on Learning Representation, 2023
  9. ICRA
    TrafficGen: Learning to Generate Diverse and Realistic Traffic Scenarios
    Lan Feng*, Quanyi Li*Zhenghao Peng*, Shuhan Tan, and Bolei Zhou
    In International Conference on Robotics and Automation, 2023
  10. NeurIPS
    Human-AI Shared Control via Policy Dissection
    Quanyi Li, Zhenghao Peng, Haibin Wu, Lan Feng, and Bolei Zhou
    In Advances in Neural Information Processing Systems, 2022
  11. ICLR
    Efficient Learning of Safe Driving Policy via Human-AI Copilot Optimization
    Quanyi Li*Zhenghao Peng*, and Bolei Zhou
    In International Conference on Learning Representations, 2022
  12. NeurIPS
    Learning to Simulate Self-driven Particles System with Coordinated Policy Optimization
    Zhenghao Peng, Quanyi Li, Chunxiao Liu, and Bolei Zhou
    In Advances in Neural Information Processing Systems, 2021
  13. CoRL
    Safe Driving via Expert Guided Policy Optimization
    Zhenghao Peng*, Quanyi Li*, Chunxiao Liu, and Bolei Zhou
    In 5th Annual Conference on Robot Learning , 2021
  14. TPAMI
    MetaDrive: Composing Diverse Driving Scenarios for Generalizable Reinforcement Learning
    Quanyi Li*Zhenghao Peng*, Lan Feng, Qihang Zhang, Zhenghai Xue, and Bolei Zhou
    In , 2021

Fun facts: