About Me

Hi there! I am Jiayao Li, an undergraduate student majoring in Computer Science and Technology (Elite Class) at Guohao College, Tongji University, with a GPA of 4.66/5 (equivalent to 93.2/100).

I have been granted eligibility for recommended graduate admission for the 2027 cohort and plan to apply for direct Ph.D. programs. My research interests lie in Embodied AI & Robotics, Physical-World AI & Simulation, and 3D Perception & Spatial Understanding.

I am particularly interested in how embodied agents can acquire high-quality interaction data, generalize under distribution shifts, diagnose failures, and improve through closed-loop feedback in real-world environments.

Research Interests

I study embodied agents through a closed loop of spatial perception, long-horizon reasoning, and feedback-driven improvement.

01

Perceive

Build multimodal spatial understanding from vision, language, geometry, and embodied interaction signals.

  • 3D Scene Representation
  • Multimodal Perception
  • Spatial Understanding
02

Reason

Study how embodied agents plan, decide, and generalize over long horizons in physical-world tasks.

  • Robot Learning
  • Task Planning
  • Long-horizon Decision-making
03

Improve

Diagnose failures and close the feedback loop so agents can adapt under distribution shifts.

  • Failure Diagnosis
  • Closed-loop Feedback
  • Robust Generalization

Selected Projects

Here are some selected projects. Click any card that interests you to view more details.

Selected Honors

News

  1. Received the Outstanding Winner & INFORMS Award in the 2026 MCM/ICM.
  2. Submitted a research manuscript on LLM reliability in library recommendation and code migration.
  3. Began Vision-and-Language Navigation (VLN) study, exploring cross-modal grounding, memory compression, and closed-loop error correction in long-horizon navigation tasks.
  4. Started working on an empirical study of LLM trustworthiness and reliability.
  5. Submitted a research manuscript on TwInS: Two-Stream Interactive Joint Learning of Scene Parsing and Geometric Vision Tasks.
  6. Started working on the TwInS project, focusing on scene parsing, stereo matching, optical flow, and cross-task representation learning.
  7. Conducted research training in medical image processing under faculty supervision.
  8. Entered Tongji University as an undergraduate student.