Hu Tianrun

About Me

I am a Research Engineer at the Smart System Institute (SSI), National University of Singapore, supervised by Prof. David Hsu. I work closely with Dr. Hanbo Zhang while at NUS. My research interests include robotics learning, manipulation, perception, planning, and Human Robot Interaction (HRI). Previously, I was an undergraduate in Computer Engineering at the College of Computing and Data Science (CCDS), Nanyang Technological University under the guidance of Prof. Lam Siew Kei. During my undergraduate studies, I worked with Dongshuo Zhang.

Feel free to reach out for collaboration and discussion of research ideas!

Research Interests

I study how to extract transferable skills from human demonstrations. My current focus is organized around three questions: what representation captures a skill, how to extract it reliably from noisy human data, and how to apply it to solve new tasks under uncertainty. While these questions cut across areas like HRI, TAMP, and POMDP-based decision-making, they serve as supporting tools rather than ends in themselves.

• Representation of skills and task-relevant state
  • Learning state representation from noisy observations and demonstrations
  • Object-centric skill representations capturing entities
  • Constraints extraction from demonstrations to recover task, safety, and kinematic constraints
• Extraction of skill representations from human data
  • Concept learning of task-relevant entities, relations, affordances, and constraints
  • Correspondence matching from demonstrations to novel scenes to ground the representation
• Application of learned skills to new tasks
  • Using Learned policy to apply skill representations across tasks and environments
  • Using TAMP to sequence skills and motion primitives
  • POMDP-style reasoning for decision-making under uncertainty

My long-term goal is to build skill-centric, demonstration-aware planning systems that are robust to real-world uncertainty—robots that learn from people, represent what they learn compactly, and reuse those skills reliably in dynamic environments.