Research

• Skill-based learning extends the capability of robot learning to tasks with a much longer horizon via informed exploration and temporally abstracted policy learning. [SkiMo 2022, SPiRL 2020]
• Learning skills and world models from unsupervised interactions or diverse unlabeled data can lead to fast learning of a new task. [TLDR 2024, DreamSmooth 2024]
• Robot learning in the real world requires high sample efficiency, offline training, and safe exploration. This can be accelerated with the extensive use of large task-agnostic data and simulators. [LEQ 2024, DROID 2024, Open X-Embodiment 2024]

Funding

We are now on track to fund our research :)

• 

  Yonsei University Center for Innovative AGI Education and Research

  Brain Korea 21 (BK21)

  $60k for 2.5 years

  Mar 2025 - Aug 2027

• 

  Fundamental AI Algorithm Research

  Global AI Frontier Lab

  $100k for 4.5 years

  Jul 2024 - Dec 2028

• 

  Hierarchical World Models for Embodied Intelligence

  AI Research Center

  $100k for 4.5 years

  Jul 2024 - Dec 2028

• 

  Learning General-Purpose Humanoid Robots via Robot Abstraction and Active 3D Perception

  NRF Outstanding Young Scientist

  $200k for 5 years

  Apr 2024 - Mar 2029

• 

  Generalizable Robotic Manipulation

  ETRI

  $40k

  Apr 2024 - Oct 2024

• 

  Skill-based Reinforcement Learning for Mobile Bimanual Manipulation

  LG Electronics

  $70k

  Mar 2024 - Dec 2024

• 

  AI Graduate School Program

  Yonsei University / IITP

  $40K for 6 years

  Mar 2024 - Feb 2030

• 

  Startup Funding

  Yonsei University Future-Leading Research Initiative

  $40K for 3 years

  May 2024 - Apr 2027