Jiho Lee

In my Ph.D. dissertation, I investigate how robots can continually improve their behavior by learning from useful experience (Active Knowledge Acquisition), unlearning harmful behavior (Selective Knowledge Removal), verifying uncertain behavior (Self-corrective Knowledge Refinement), and grounding their behavior to the current physical context (Environment-aware Knowledge Alignment). Title Toward Adaptive Robot Learning with Foundation Models: From Task Planning to Action Policies

My research experience lies in adaptive learning inspired by how humans learn and adapt to the world, spanning data construction, training, and deployment, with an emphasis on data reliability and resource efficiency to enable models to continuously self-improve while making their behavior more reliable, transferable, and context-aware.