Jiwei Yang

Jiwei Yang 杨济玮

Research Assistant

Tsinghua University

yangjiwei@tsinghua.vin

I am a research assistant at the School of Vehicle and Mobility, Tsinghua University, advised by Prof. Jun Li, working on world models.

I received my bachelor’s degree in Software Engineering from Guangzhou Institute of Science and Technology in June 2026 and completed my joint training at the School of Vehicle and Mobility, Tsinghua University. Previously, I worked as an embodied intelligence algorithm engineer at the Center for Brain Inspired Computing Research (CBICR), Tsinghua University.

Research Interests

My research focuses on learning and reasoning about the real world with world models, particularly real-world dynamics and interactions: understanding how the world evolves and how agents interact, beyond predicting the next observation.

News

Education

Publications

* equal contribution † corresponding author

Show publications
  1. Auto-JEPA: A Latent World Model of Continuous Intent for End-to-End Autonomous Driving teaser
    Auto-JEPA: A Latent World Model of Continuous Intent for End-to-End Autonomous Driving
    Jiwei Yang, Zhengxian Chen†, Chaoshen Huang, Jun Li
    🌟 Cited by Yann LeCun.
    ArXiv
    arXiv PDF Code
  2. SDRA: Spatiotemporally Decoupled VLM-Enhanced Reinforcement Learning for Energy-Efficient Heavy-Vehicle Driving teaser
    SDRA: Spatiotemporally Decoupled VLM-Enhanced Reinforcement Learning for Energy-Efficient Heavy-Vehicle Driving
    Jiwei Yang*, Zhengxian Chen*†, Pengyi Xiao, Weihua Pan, Chenggang Zhang, Lei Yang, Wenfeng Guo, Chaoshen Huang, Jun Li
  3. OP-RTDETR: Real-Time Outdoor Parking Space Detection with Grouped Convolutions and Group-Level Contrastive Denoising teaser
    OP-RTDETR: Real-Time Outdoor Parking Space Detection with Grouped Convolutions and Group-Level Contrastive Denoising
    Jiwei Yang, Huilin Chen, Weihua Pan, Ting Liu, Si Li, Yujie Liu, Yingrun Li
  4. CHyFN: A Consistency-Aware Hybrid Feature Network for Robust Object Recognition in Complex Scenes teaser
    CHyFN: A Consistency-Aware Hybrid Feature Network for Robust Object Recognition in Complex Scenes
    Shuo Ye, Jiwei Yang, Zitong Yu, Dan Guo, Shutao Xia, Richang Hong, Meng Wang, Fellow, IEEE
  1. Auto-JEPA: A Latent World Model of Continuous Intent for End-to-End Autonomous Driving teaser
    Auto-JEPA: A Latent World Model of Continuous Intent for End-to-End Autonomous Driving
    Jiwei Yang, Zhengxian Chen†, Chaoshen Huang, Jun Li
    🌟 Cited by Yann LeCun.
    ArXiv
    arXiv PDF Code
  2. SDRA: Spatiotemporally Decoupled VLM-Enhanced Reinforcement Learning for Energy-Efficient Heavy-Vehicle Driving teaser
    SDRA: Spatiotemporally Decoupled VLM-Enhanced Reinforcement Learning for Energy-Efficient Heavy-Vehicle Driving
    Jiwei Yang*, Zhengxian Chen*†, Pengyi Xiao, Weihua Pan, Chenggang Zhang, Lei Yang, Wenfeng Guo, Chaoshen Huang, Jun Li
  3. OP-RTDETR: Real-Time Outdoor Parking Space Detection with Grouped Convolutions and Group-Level Contrastive Denoising teaser
    OP-RTDETR: Real-Time Outdoor Parking Space Detection with Grouped Convolutions and Group-Level Contrastive Denoising
    Jiwei Yang, Huilin Chen, Weihua Pan, Ting Liu, Si Li, Yujie Liu, Yingrun Li
  4. CHyFN: A Consistency-Aware Hybrid Feature Network for Robust Object Recognition in Complex Scenes teaser
    CHyFN: A Consistency-Aware Hybrid Feature Network for Robust Object Recognition in Complex Scenes
    Shuo Ye, Jiwei Yang, Zitong Yu, Dan Guo, Shutao Xia, Richang Hong, Meng Wang, Fellow, IEEE

Selected Projects

  1. Auto-JEPA training and inference architecture
    Auto-JEPA 🔥
    A latent-space world model for learning continuous driving intent from visual observations, ego-motion history, and navigation commands.
    Code

Experience

Academic Experience

  1. Tsinghua University
    School of Vehicle and Mobility, Tsinghua University
    Research Assistant: Jun. 2026 - Present.
    Advisor: Prof. Jun Li.
    Topics: latent world models and autonomous driving.
  2. Tsinghua University
    Center for Brain Inspired Computing Research (CBICR), Tsinghua University
    Embodied Intelligence Algorithm Engineer: Aug. 2025 - Oct. 2025.
    Topics: Multimodal perception, language understanding, and action planning.

Honors & Awards