I build embodied agents that can continually acquire, reuse, and compose knowledge for complex robotic tasks. My work brings together lifelong learning, reinforcement learning, language-guided reasoning, and robot manipulation.
Project descriptions and links
- LEGION — Lifelong reinforcement learning with a Bayesian non-parametric knowledge space and language embeddings.
- DAHLIA — Progressive in-context code generation and episodic visual feedback for long-horizon manipulation.
- HELIOS — Hierarchical reinforcement learning with adaptive Bayesian non-parametric skill priors.
- LYRA — Human-guided lifelong code generation for reusable and extensible robotic skills.
- ReMAP — Transfer of reusable meta-knowledge across agents with different embodiments.
Preserving and combining knowledge in robotic lifelong reinforcement learning
Nature Machine Intelligence, 7, 256–269 (2025)
Paper ·
Project page ·
Code